<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Johannes Keizer on Medium]]></title>
        <description><![CDATA[Stories by Johannes Keizer on Medium]]></description>
        <link>https://medium.com/@johannes-keizer?source=rss-3bdd1ad4f6f0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*z8qhTaM3etJ42yw1.jpg</url>
            <title>Stories by Johannes Keizer on Medium</title>
            <link>https://medium.com/@johannes-keizer?source=rss-3bdd1ad4f6f0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 21:53:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@johannes-keizer/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[An ontology that asks to be corrected — new on pd.sciai.eu]]></title>
            <link>https://johannes-keizer.medium.com/an-ontology-that-asks-to-be-corrected-new-on-pd-sciai-eu-8ed2150a53fe?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/8ed2150a53fe</guid>
            <category><![CDATA[parkinsons-disease]]></category>
            <category><![CDATA[ontology]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Thu, 01 Oct 2026 02:57:11 GMT</pubDate>
            <atom:updated>2026-10-01T02:57:11.180Z</atom:updated>
            <content:encoded><![CDATA[<p>300,000 entity strings, five classes of errors, and an open invitation: help us fix the Parkinson&#39;s disease ontology.</p><p>PD.sciai.eu/blog/ontology_crowdsourcimg_error_20260929</p><p>Cross-post note: the full article lives on our results site.</p><p>The newest article on pd.sciai.eu is not a success report — it is a catalogue of what went wrong when we normalized 300,000 entity strings from the Parkinson’s literature, and an invitation to help fix it:</p><p><a href="https://pd.sciai.eu/blog/ontology_crowdsourcing_errors_20260929/">An ontology that asks to be corrected</a></p><p>Four spelling variants of “L-DOPA-induced dyskinesia” hiding 22 papers of evidence as separate concepts. 142 canonical entities that were all “Hoehn &amp; Yahr”. A MeSH descriptor for the emotion “Rage” accumulating synonyms for the receptor RAGE. The article walks through the five failure classes we found, states our curation rules (automation with safety thresholds; a human for every case of doubt; every decision rebuild-proof), and explains why the Parkinson’s disease ontology (sciai-lab) is published early, with a DOI: so that its errors can be found — by domain experts, in small, well-prepared units of work.</p><p>If you know a biocurator or PD researcher who enjoys exactly this kind of looking — send them the link.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8ed2150a53fe" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Same Question, Two Answers: What Evidence Discipline Adds]]></title>
            <link>https://johannes-keizer.medium.com/same-question-two-answers-what-evidence-discipline-adds-175c2114a19a?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/175c2114a19a</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[glp-1]]></category>
            <category><![CDATA[parkinson-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Sat, 19 Sep 2026 12:19:51 GMT</pubDate>
            <atom:updated>2026-09-19T12:19:51.103Z</atom:updated>
            <content:encoded><![CDATA[<h3><em>A live experiment: we gave one research question to a frontier AI model as a bare chatbot prompt, and to the sciai pipeline as a Computer assisted Research Analysis (CRA). Same question, same day. Here is what methodology adds.</em></h3><h3>The question</h3><blockquote><strong><em>Do GLP-1 receptor agonists modify the progression of Parkinson’s disease? Summarize the current clinical and mechanistic evidence, where it contradicts itself, and what the open questions are.</em></strong></blockquote><p>GLP-1 receptor agonists (the drug class behind Ozempic and Wegovy) are one of the most watched candidate treatments in Parkinson’s research — an ideal test question, because the evidence is large, recent, and genuinely contradictory.</p><h3>The two contestants</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6PHbZtcNghesVFdQWbZgVA.png" /></figure><h3>The answers</h3><p><strong>The chatbot</strong> answered competently — it knows the four big trials (exenatide phase 2 and 3, LIXIPARK, NLY01) and judges them sensibly:</p><blockquote><em>“GLP-1 receptor agonists remain plausible experimental treatments for Parkinson’s disease, but they have not been shown convincingly to slow its underlying progression. … </em>This summary includes the pivotal published results through 2025. I do not have live literature access to verify subsequent trial readouts.<em>”</em></blockquote><p>A good answer — for 2025. It cites four trials from memory, with author names but no verifiable references, and it cannot know what it missed.</p><p><strong>The CRA</strong> answered from the current literature — 256 papers up to 2026, every claim carrying a checkable citation:</p><blockquote><em>“GLP-1RAs demonstrate robust, reproducible neuroprotective effects across diverse preclinical PD models… However, translation to clinical disease modification remains unproven and increasingly contested. … the pivotal phase III Exenatide-PD3 trial (n=194) found no significant benefit on motor progression over 96 weeks (PMID: 39919773)… Meta-analyses of RCTs yield conflicting results (PMID: 39642803, 40288416, 42105060)…”</em></blockquote><p>Same bottom line — but every clause is anchored to a specific paper you can open and check.</p><h3>What the methodology adds</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KrFXfLwJqfVbbQXl3JQqEQ.png" /></figure><h3>Why this matters</h3><p>The interesting finding is <em>not</em> that the chatbot is wrong — it isn’t. Even a good frontier answer covers what the model happens to remember, and nothing it says can be checked without redoing the work. The gap is not intelligence; it is <strong>evidence discipline</strong>: defined corpus, structured extraction, measured coverage, checkable citations, independent audit.</p><p>That discipline is what makes an AI answer usable for science.</p><h3>See for yourself</h3><ul><li><a href="https://sciai.eu/downloads/astra_chatbot_glp1_20260919.md">The full chatbot answer (raw, unedited)</a></li><li><a href="https://sciai.eu/downloads/SciAI-CRA_20260919_glp1-parkinsons_v1.6.0.md">The full CRA report (markdown)</a></li><li><a href="https://pd.sciai.eu/kg/">Explore the knowledge graph behind it (playground, on pd.sciai.eu)</a></li><li>The ontology and graph data are published openly on Zenodo (CC BY-NC 4.0): <a href="https://doi.org/10.5281/zenodo.22759155">doi:10.5281/zenodo.22759155</a></li></ul><p><em>SciAI CRA pipeline v1.6.0 · synthesis: Kimi K3 · audit: DeepSeek · extraction: universal_pd schemas over 256 papers · chatbot: OpenAI GPT-6 Astra via OpenRouter, single bare prompt, 2026–09–19. The CRA text is published CC BY 4.0.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=175c2114a19a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using SciAI — small questions, big questions]]></title>
            <link>https://johannes-keizer.medium.com/using-sciai-small-questions-big-questions-77bf61c18166?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/77bf61c18166</guid>
            <category><![CDATA[sleep-disturbances]]></category>
            <category><![CDATA[melatonin]]></category>
            <category><![CDATA[parkinsons-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Sat, 18 Jul 2026 07:50:29 GMT</pubDate>
            <atom:updated>2026-07-18T07:50:29.254Z</atom:updated>
            <content:encoded><![CDATA[<p>While studying sleep disturbances in Parkinson’s disease (PD), I first produced exhaustive material on pramipexole and then turned to melatonin, an agent discussed not only for improving sleep quality but also for possible neuroprotective effects.</p><h3>Example 1: The landscape question (high-recall strategy)</h3><p>My first research question was deliberately broad: <em>“Melatonin Effects on Parkinson’s Disease-Related Sleep Disturbances.”</em> The search strategy was exploratory and designed for high recall. I cast a wide semantic net that included melatonin, circadian rhythm, PD, sleep architecture, REM sleep behavior disorder, and neuroprotection. Because the goal was a comprehensive landscape review, I allowed the corpus to grow to 88 papers before running the CRA synthesis.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JWJZ45n9wbqLoR1IjX44jQ.png" /></figure><p>picture produced by Notebook LM (Google)</p><blockquote>Executive summary:<em> Sleep disturbances represent one of the most prevalent and clinically significant non-motor symptoms of Parkinson’s disease (PD), affecting 64–90% of patients depending on disease stage and assessment methodology (PMID: 10704535; PMID: 20187236; PMID: 28261151; PMID: 39508600). These disturbances frequently manifest years before motor symptom onset, positioning them as potential prodromal markers of synucleinopathy (PMID: 10704535; PMID: 31807917; PMID: 41467259). The clinical consequences extend far beyond nighttime discomfort: sleep disorders independently predict poor quality of life (PMID: 20187236; PMID: 27863901; PMID: 28802939), exacerbate depressive symptoms, elevate mortality rates, and intensify both motor and non-motor symptom burden (PMID: 37620177). Despite this substantial disease burden, sleep problems remain frequently underrecognized and ineffectively treated in clinical practice (PMID: 15994219).</em></blockquote><blockquote><em>Melatonin has emerged as a particularly promising therapeutic candidate spanning symptomatic sleep improvement and potential disease modification. The evidence base has evolved considerably across three decades of research, from early observational studies of circadian dysfunction through preclinical neuroprotective investigations to contemporary meta-analyses of randomized controlled trials. This meta-synthesis integrates findings from 1989–2026 across reviews, mechanistic studies, clinical trials, and biomarker investigations to provide a comprehensive assessment of melatonin’s effects on PD-related sleep disturbances.</em></blockquote><blockquote><em>Key established findings include: melatonin system dysfunction is intrinsic to PD pathophysiology, encompassing reduced rhythmic amplitude, receptor downregulation in sleep-regulating and dopaminergic structures, and pineal gland calcification (PMID: 22010042; PMID: 33935759; PMID: 41143249; PMID: 38259379); exogenous melatonin significantly improves subjective sleep quality across multiple meta-analyses, with pooled effect sizes consistently favoring active treatment (PMID: 35185525; PMID: 37881313; PMID: 40288252; PMID: 40387966; PMID: 40588412); immediate-release formulations and shorter treatment durations (4 weeks) appear more effective than prolonged-release or longer courses (PMID: 40288252); melatonin exhibits multifaceted neuroprotective mechanisms in preclinical models, including antioxidant, anti-inflammatory, anti-apoptotic, autophagy-modulating, mitochondrial-protective, and α-synuclein-inhibiting effects (PMID: 27829983; PMID: 36829955; PMID: 37881313; PMID: 40068276; PMID: 41594924); and a bidirectional relationship exists whereby neurodegeneration promotes sleep disruption, which in turn accelerates pathogenesis through neuroinflammation, glymphatic dysfunction, and toxic protein accumulation (PMID: 31807917; PMID: 35938257; PMID: 36361716; PMID: 39508600).</em></blockquote><blockquote><em>Critical unresolved tensions permeate the evidence base. Endogenous melatonin levels are paradoxically elevated in some PD cohorts rather than deficient (PMID: 32508583), challenging simple replacement models. Effects on objective polysomnographic measures remain inconsistent despite consistent subjective improvements (PMID: 17404779; PMID: 40387966; PMID: 40588412). Motor symptom benefits are suggested by one meta-analysis (PMID: 37881313) but refuted by others (PMID: 40387966). The translation from compelling preclinical neuroprotection to demonstrated human disease modification remains entirely unproven. Perhaps most fundamentally, non-melatonin interventions such as transcranial direct current stimulation can improve sleep while reducing melatonin levels (PMID: 34917270), suggesting that circadian resynchronization rather than absolute hormone levels may underlie therapeutic effects.</em></blockquote><blockquote><em>The clinical positioning of melatonin is further complicated by its favorable safety profile compared to alternatives — melatonin matched clonazepam and trazodone for global sleep improvement with superior tolerability and RBD symptom reduction in head-to-head comparison (PMID: 35660990) — yet uncertainty regarding optimal dosing (2–50 mg studied), formulation (immediate- versus prolonged-release), and treatment duration. Most reviewers across all periods converge on the need for larger, well-designed randomized controlled trials with objective endpoints, longer follow-up, and explicit assessment of disease-modifying potential (PMID: 33935759; PMID: 40387966; PMID: 41594924).</em></blockquote><p>Full article: <a href="https://pd.sciai.eu/library/cras/sciai-cra_20260714_has-melatonin-effect-on-pd-related-sleep-disturban/">https://pd.sciai.eu/library/cras/sciai-cra_20260714_has-melatonin-effect-on-pd-related-sleep-disturban/</a></p><h3>Example 2: The pinpoint question (high-precision strategy)</h3><p>The first synthesis covered mechanisms and subjective outcomes but left a specific clinical gap: dosing and safety. For the second query, I shifted from an exploratory to a surgical search strategy. The question was narrow: <em>“What are the doses of melatonin used to treat PD sleep disturbances, and from which dose onward do side effects emerge?”</em> I constrained the search with targeted keyword combinations — explicitly favoring terms such as “mg,” “dose,” “adverse effects,” and “safety” alongside “melatonin” and “Parkinson” — to capture only papers reporting dose–response or tolerability data. This high-precision approach yielded a compact corpus of just 16 PMIDs.</p><blockquote>Executive summary:<em> This meta-synthesis integrates evidence from systematic reviews, meta-analyses, and primary clinical studies to address the research question: What are the doses of melatonin used to treat PD sleep disturbances, and from which dose onward do side effects emerge?</em></blockquote><blockquote><em>The evidence demonstrates that melatonin improves subjective sleep quality in Parkinson’s disease (PD) across a wide dose range (2–50 mg), with the most robust and replicated efficacy at 3–5 mg for general sleep disturbances and ≥10 mg for combined sleep and motor benefits. However, a critical finding emerges: no study or systematic review establishes a definitive threshold dose for adverse effects in PD populations. The lowest dose at which specific adverse effects (morning headaches, morning sleepiness, delusions/hallucinations) have been reported is 3 mg in a retrospective, uncontrolled series (PMID: 14592300), but this evidence is weak. The highest dose tested in PD, 50 mg, showed no reported adverse effects in its single study (PMID: 16084125), though this likely reflects incomplete reporting rather than true absence of harm.</em></blockquote><blockquote><em>The evidence base is characterized by: (a) consistent efficacy for subjective sleep outcomes at low-to-moderate doses; (b) suggestive but less certain benefits at higher doses; © a fundamental and persistent gap in dose-specific safety characterization; and (d) contradictory findings regarding optimal formulation and duration that preclude definitive clinical recommendations.</em></blockquote><p>Full article: <a href="https://pd.sciai.eu/library/cras/sciai-cra_20260716_what-are-the-doses-of-melatonin-used-to-treat-pd-sleep/">https://pd.sciai.eu/library/cras/sciai-cra_20260716_what-are-the-doses-of-melatonin-used-to-treat-pd-sleep/</a></p><h3>The lesson: match strategy to scope</h3><p>The difference in corpus size was not incidental; it reflected a deliberate trade-off between recall and precision. The broad question required a high-recall strategy to map an entire field (88 papers). The narrow question demanded high precision to isolate a specific clinical variable (16 papers). Ten of those sixteen were already inside the first corpus — confirmation that the wide net had captured them — but six were new, retrieved only because the targeted query privileged dosing and safety terminology over general sleep discourse. In other words, the focused strategy found needles that the broad sweep had missed.</p><p>From SciAI’s repository of roughly 200,000 Parkinson papers, you can paint a sweeping knowledge landscape or target a precise data point. The key is calibrating your search strategy to the scope of the question.</p><p>Important:</p><ul><li>SciAI is research analysis, NO medical advice</li><li>Big parts of SciAI outputs are produced by LLM. LLMs like humans can make errors</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=77bf61c18166" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[# SciAI: How I’m Turning  180,000  Parkinson’s Papers Into Structured Knowledge]]></title>
            <link>https://johannes-keizer.medium.com/sciai-how-im-turning-half-a-million-parkinson-s-papers-into-structured-knowledge-5238c8bf9b4b?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/5238c8bf9b4b</guid>
            <category><![CDATA[llm-applications]]></category>
            <category><![CDATA[knowledge-graph]]></category>
            <category><![CDATA[parkinson-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Wed, 08 Jul 2026 14:47:09 GMT</pubDate>
            <atom:updated>2026-10-05T03:58:11.832Z</atom:updated>
            <content:encoded><![CDATA[<p><strong># SciAI: How I’m Turning 180,000 Parkinson’s Papers Into Structured Knowledge</strong></p><p>— -</p><p><strong>Why I started this</strong></p><p>I was diagnosed with Parkinson’s disease in 2021. I am still convinced that the time has come to defeat it — not someday, but now.</p><p>I finished my PhD in 1992, when the growth of scientific production was still roughly linear. Since 2000 it has become exponential. Today, PubMed adds more than a million new biomedical papers every year. For a researcher studying Parkinson’s disease (PD), this is not a library — it is a flood. A query for “Parkinson’s disease” returns hundreds of thousands of results. A focused sub-query — say, “proton pump inhibitors and Parkinson’s disease” — still leaves thousands of candidates to read, classify, and synthesize. And most of that information is trapped in scientific prose, not structured data.</p><p>I spent my working life in science ([johanneskeizer.com/cv](https://johanneskeizer.com/cv)) — 15 years in the lab, 25 years in data crunching. What happened in AI during the last three years excited me too much to stay on the sidelines. I had to dig in myself.</p><p>Back in the early 2000s, when I was still in scientific documentation, I wanted to change scientific publishing into a network of semantic statements — statements that could be combined, compared, and organized into ontologies. The technology was not ready then. Now it is. That is why I built <strong>**SciAI**</strong>.</p><p>SciAI is not a glorified chatbot. It is a knowledge infrastructure for scientific literature, currently focused on <strong>**Parkinson’s disease**</strong>. At its core are three overlapping representations of the same corpus — what I call the <strong>**three knowledge layers**</strong> — and a pipeline that turns raw papers into structured, verifiable, citable evidence.</p><p>This article explains why that layered architecture matters, why PD is the perfect proving ground, and what I have learned about working with LLMs — including what running local large language models is really like once the hype wears off.</p><p>— -</p><p><strong>The problem: science is drowning in its own output</strong></p><p>Search engines solve the <em>*retrieval*</em> problem. They do not solve the <em>*understanding*</em> problem. A list of 3,000 paper titles is not a review. A PDF full-text is not a structured finding. And a ChatGPT answer that hallucinates a citation is worse than useless in science.</p><p>What researchers need is not another answer engine. They need a system that reads the literature once, structures it, and lets them query it forever.</p><p>— -</p><p><strong>Layer 1: MeSH — the curated map</strong></p><p>The first layer is <strong>**MeSH**</strong> (Medical Subject Headings), the National Library of Medicine’s controlled vocabulary. Every paper in PubMed is annotated with MeSH terms: diseases, chemicals, anatomical structures, study types, and more.</p><p>MeSH is the closest thing biomedicine has to a shared ontology. It is stable, authoritative, and computable. In SciAI, MeSH serves three purposes:</p><p>1. <strong>Filtering.**</strong> A user can say, “Show me only papers that mention levodopa <em>*and*</em> dyskinesia,” and MeSH makes that precise.</p><p>2. <strong>Navigation.**</strong> MeSH terms become nodes in a knowledge graph, letting a researcher walk from “Parkinson’s disease” to “alpha-synuclein” to “mitochondrial dysfunction.”</p><p>3. <strong>Grounding.**</strong> Because MeSH is curated, it anchors vaguer queries in established terminology.</p><p>But MeSH has limits. It is shallow: a paper tagged with “Parkinson Disease” does not tell you whether it is about genetics, biomarkers, or deep brain stimulation. It is also slow to evolve. New mechanisms, drug candidates, and methodological concepts appear in papers long before they become official headings.</p><p>MeSH gives you the map. It does not give you the terrain.</p><p>— -</p><p><strong>Layer 2: Vectors — the semantic sea</strong></p><p>The second layer is <strong>**dense vector embeddings**</strong>. I embed paper titles, abstracts, and — where available — full texts into a high-dimensional semantic space. Papers about similar ideas cluster together even if they use different keywords.</p><p>Vector search is what makes SciAI feel almost telepathic. A user can type a natural-language research question like:</p><p>&gt; “What is the evidence that gut microbiome changes precede motor symptoms in Parkinson’s disease?”</p><p>Even if no paper uses that exact phrasing, vector search surfaces the relevant literature because the embeddings capture meaning, not just term frequency.</p><p>In SciAI, vectors operate at two scales:</p><p>- <strong>**Document-level embeddings**</strong> rank and retrieve candidate papers.</p><p>- <strong>**Chunk-level embeddings**</strong> let the system find specific passages inside papers — the paragraph that actually describes a method, a result, or a limitation.</p><p>Vectors are powerful, but they are also opaque. A vector tells you that two papers are <em>*similar*</em>; it does not tell you <em>*why*</em>, or what the evidence actually says. Two papers may cluster together because they share a method, a disease, or a buzzword — not because they agree on a finding.</p><p>That is why vectors alone are not enough.</p><p>— -</p><p><strong>Layer 3: Extractions — the structured facts</strong></p><p>The third layer is what makes SciAI different from a semantic search engine or a retrieval-augmented chatbot. I run every paper through a <strong>**structured extraction schema**</strong> and convert free text into machine-readable fields:</p><p>- Study type and methodology</p><p>- Sample size and population</p><p>- Key findings and direction of effect</p><p>- Mechanisms and pathways</p><p>- Compounds, proteins, genes, and species</p><p>- Evidence quality and limitations</p><p>- Relevance to the research question</p><p>These extractions are stored in a relational database as typed JSON. They are not summaries. They are <em>*structured records*</em> that can be counted, filtered, grouped, and synthesized.</p><p>This is the layer where scientific literature becomes data.</p><p>Here is why it matters in practice. A traditional systematic review might take a team six months to screen, extract, and synthesize a few hundred papers. In SciAI, the extraction layer lets a single researcher:</p><p>- Build a corpus of thousands of papers in hours</p><p>- See how many studies are clinical trials versus animal models versus reviews</p><p>- Group findings by mechanism (e.g., “anti-inflammatory,” “α-synuclein clearance,” “mitochondrial protection”)</p><p>- Generate a narrative synthesis with citations linked back to specific PMIDs</p><p>The output is still human-reviewed. The machine does the tedious work; the researcher does the thinking.</p><p>— -</p><p><strong>## The three layers together: why 1 + 1 + 1 &gt; 3</strong></p><p>Each layer compensates for the weaknesses of the others.</p><p>| Layer | Strength | Weakness | Compensated by |</p><p>| — — — -| — — — — — | — — — — — | — — — — — — — — |</p><p>| MeSH | Precise, curated, authoritative | Shallow, slow to update | Vectors + Extractions |</p><p>| Vectors | Captures semantic similarity | Opaque, no structured facts | Extractions + MeSH |</p><p>| Extractions | Structured, verifiable, countable | Depends on extraction quality | MeSH + Vectors |</p><p>MeSH finds the neighborhood. Vectors find the conversation. Extractions extract the claims. Together, they let a researcher move from “find me papers” to “tell me what the evidence says.”</p><p>I believe this layered architecture is the right way to build scientific AI. A chatbot that reads abstracts on demand is a demo. A system that maintains persistent, structured, queryable knowledge across millions of papers is infrastructure.</p><p>— -</p><p><strong>## Why Parkinson’s disease?</strong></p><p>SciAI is not generic — at least not yet. I have concentrated on <strong>**Parkinson’s disease**</strong> for a deeply personal reason, but also for a strategic one: focus beats breadth when you are trying to build something real.</p><p>PD is appropriate for three reasons.</p><p><strong>**First, the literature is large but bounded.**</strong> There are roughly half a million PD-related papers in PubMed. That is enough to stress-test every part of the pipeline — search, ranking, extraction, synthesis — but small enough that a focused effort can still validate the outputs against domain expertise.</p><p><strong>**Second, the research landscape is rich and heterogeneous.**</strong> PD spans genetics (LRRK2, GBA1, SNCA), molecular mechanisms (alpha-synuclein, mitochondrial dysfunction, neuroinflammation), clinical phenotypes (tremor, gait, cognition, sleep), pharmacology (levodopa, dopamine agonists, MAO-B inhibitors), and non-pharmacological interventions (exercise, DBS, focused ultrasound). No single researcher can hold all of this in memory. A structured knowledge layer becomes genuinely useful.</p><p><strong>**Third, the patient and research communities are engaged and underserved.**</strong> Despite decades of research, disease-modifying therapies remain elusive. Researchers need better tools to see patterns across sub-fields, identify contradictions, and avoid redundant studies. I believe a PD-focused knowledge graph can accelerate that work.</p><p>The result is what I think may be one of the most comprehensive structured representations of Parkinson’s disease literature in existence: tens of thousands of papers, millions of extracted facts, and a growing network of mechanisms, compounds, and evidence types.</p><p>— -</p><p><strong>## I had no plan, only an idea — and LLMs were essential</strong></p><p>When I started this, I had only a vague idea. I did not know about vectors, RAG, or multi-modal models. I began talking about my idea of a “Parkinson information system” with ChatGPT. It encouraged me, as LLMs do.</p><p>I learned Python by building. I made many experiments — some of them published on Medium. I started from scratch at least four times.</p><p>I abandoned OpenAI because they abandoned me while I was working from Wuhan. Claude could be reached with a good VPN, and I worked with Claude for a long time. At a certain point I did not want to hear Dario Amodei’s carefully calibrated certainty every day. I was again in China, and I switched to <strong>**Kimi**</strong>. Kimi and I built the actual version of SciAI together. In the beginning Kimi was messier than Claude, but I felt more comfortable with it. With Kimi k1.5 / Kimi Code, I do not miss Claude anymore.</p><p>I keep an Anthropic account for very complex reports — they are still the best at that — but for normal development I am with Kimi. For bulk extraction work, I use <strong>**DeepSeek**</strong>. A recent 500-paper DeepSeek extraction cost about <strong>**$0.69**</strong>. Anthropic would have cost roughly <strong>**$15**</strong> for more or less the same result.</p><p>At the end of the day, I do not care about frontier benchmarks. I care that my work gets done properly.</p><p>— -</p><p><strong>## Local LLMs: the hype and the hangover</strong></p><p>No article about AI in 2026 would be complete without addressing <strong>**local large language models**</strong>. The promise is seductive: run Llama, Qwen, or Mistral on your own GPU, pay no API fees, keep your data private, and escape rate limits.</p><p>I have spent quite a lot of time running local LLMs for extraction, synthesis, and knowledge-graph construction. They are not useless — I still use them for <strong>**relevance screening**</strong> and <strong>**entity extraction**</strong>. But the broader promise needs a reality check.</p><p><strong>### 1. Hardware is not the only cost</strong></p><p>A quantized 14B-parameter model like Qwen2.5–14B-Instruct-AWQ fits on a single consumer GPU. But “fits” is not the same as “runs well.” To extract structured JSON from thousands of full-text papers, you need:</p><p>- Enough VRAM for the model plus context (full texts are long)</p><p>- Fast inference throughput (otherwise a 10K-paper batch takes weeks)</p><p>- Reliable uptime (a crashed vLLM server breaks an overnight batch)</p><p>- Cooling and power (a GPU running at 100% for days is not silent)</p><p>The capital cost of a GPU is real. The operational cost — babysitting the server, restarting crashed processes, tuning batch sizes, managing CUDA versions — is often higher.</p><p><strong>### 2. Speed is the silent killer</strong></p><p>DeepSeek, Claude, or GPT-4 can extract a paper in a few seconds. A local 14B model on a single GPU takes 10–20 seconds per paper. That sounds fine until you multiply by 10,000 papers. Suddenly your “free” local run takes 40 hours.</p><p>And that assumes no failures. Local models are more brittle. A long paper may exceed the context window. A tricky prompt may produce malformed JSON. A driver update may break the inference server. Each failure requires human intervention or retry logic.</p><p><strong>### 3. Quality is not uniform</strong></p><p>Local models are genuinely impressive for their size. A well-quantized 14B model can produce coherent summaries and decent structured extractions. But they struggle with:</p><p>- <strong>**Nuanced reasoning**</strong> across long contexts</p><p>- <strong>**Consistent schema adherence**</strong> (e.g., always populating optional fields correctly)</p><p>- <strong>**Subtle factual distinctions**</strong> (e.g., drug-induced parkinsonism vs. idiopathic PD)</p><p>- <strong>**Citations**</strong> (local models are no less prone to hallucination than remote ones)</p><p>For high-stakes scientific extraction, the best local models today are roughly comparable to mid-tier API models from two years ago. That is remarkable. It is not enough if you need production-grade reliability.</p><p><strong>### 4. The economics are closer than they look</strong></p><p>My recent 500-paper DeepSeek test cost about <strong>**$0.69**</strong>. Extrapolated, 10,000 papers cost roughly <strong>**$14**</strong>. Even if I split half my volume to a remote provider, the total spend is trivial compared to the cost of a GPU, electricity, and the engineering time required to keep a local stack running.</p><p>Local LLMs are not dead. They are excellent for prototyping, for sensitive data, for relevance screening, for entity extraction, and for tasks where latency matters more than throughput. But for large-scale structured extraction, the “free local model” narrative is misleading. The bill just shows up as hardware, electricity, and lost sleep instead of an API invoice.</p><p>— -</p><p><strong>## Where SciAI is going</strong></p><p>SciAI today is a working research platform, not a slide deck. Researchers can:</p><p>- Build focused corpora using keyword, semantic, and MeSH filters</p><p>- Extract structured evidence from full texts</p><p>- Generate systematic-review-style reports with cited PMIDs</p><p>- Explore a growing knowledge graph of PD mechanisms and findings</p><p>My near-term roadmap has three priorities.</p><p><strong>**1. Expand the extraction coverage.**</strong> I am scaling from thousands to tens of thousands of extracted papers, using a hybrid of local vLLM and remote DeepSeek depending on cost, speed, and time-of-day rate limits.</p><p><strong>**2. Open the topic guard.**</strong> SciAI’s architecture is already topic-agnostic. I am adding per-user topic controls so researchers can apply the same pipeline to other neurodegenerative diseases — ALS, Alzheimer’s, Huntington’s — while keeping the PD corpus as the foundational layer.</p><p><strong>**3. Make the knowledge graph queryable.**</strong> Extractions are useful for synthesis; a graph is useful for discovery. I am connecting extracted entities (genes, proteins, compounds, pathways) into a navigable network that can answer questions like, “Which compounds modulate both alpha-synuclein aggregation and mitochondrial function?”</p><p>— -</p><p><strong>## A modest proposal for scientific AI</strong></p><p>The current wave of AI hype is obsessed with chat. I think that is the wrong interface for science.</p><p>A scientist does not need a chatbot that can sound confident. A scientist needs:</p><p>- A <strong>**search layer**</strong> that understands meaning, not just keywords</p><p>- A <strong>**structured layer**</strong> that turns papers into data</p><p>- A <strong>**synthesis layer**</strong> that produces verifiable, citable summaries</p><p>- A <strong>**knowledge layer**</strong> that connects findings across studies and fields</p><p>That is what I am building with SciAI. And I am starting with Parkinson’s disease because it is hard enough to be meaningful and focused enough to be tractable.</p><p>If you are a PD researcher, a computational biologist, or simply someone who believes that scientific knowledge should be structured rather than buried in PDFs, I would love to hear from you.</p><p>— -</p><p><em>*SciAI is an independent research infrastructure project. The platform is not a medical device and does not provide clinical advice. It is a tool for researchers who need to see the literature more clearly.*</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5238c8bf9b4b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Experiences with LLMs]]></title>
            <link>https://johannes-keizer.medium.com/experiences-with-llms-15a1fdaa7ef5?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/15a1fdaa7ef5</guid>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Wed, 25 Feb 2026 00:00:05 GMT</pubDate>
            <atom:updated>2026-02-26T05:02:48.612Z</atom:updated>
            <content:encoded><![CDATA[<p>I’m working now for nearly a year or so quite intensive with Chat GPT, Deep Seek and for the last months now with Claude and the APIs of their respective systems. I am a Biochemist, spent most of my professional life in data science, and I use these tools to build software for scientific research — specifically to find, acquire, analyze and synthesize literature on Parkinson’s disease. I am not a programmer and I don’t want to become one. What follows are my observations from the field.</p><p>I abandoned OpenAI because they abandoned me, when I was for some months in China. Furthermore I cannot stand their method to tell you when you are over usage quota: They don’t tell you, they simply make your life impossible by slowing down all operations. Furthermore I try to limit my business with persons who are licking the feet of Donald Trump.</p><p>I like DeepSeek, but it talks too much and was unable to install its code plugin into my VS Code. I use it at the moment only for API calls, big ones, because a process that costs 1 USD with Claude costs 1 cent with DeepSeek. But as a digital assistant, I ended up with Claude. When I realized that my Pro Subscription would not bring me where I want to go, I changed to Max and for 100 Euro/month I have a digital assistant 24/24 and 7/7 and I never arrived at the limit.</p><h3>LLM/AIs are extremely useful</h3><p>Initially I made a bit of fuss about not calling them AI, but LLMs or digital assistants, but I decided that this is stupid as Intelligence is anyway badly defined. When I hear human intelligence as a benchmark for AGI, then I get laughing. As you see from the governments they vote for, a big part of human beings is simply stupid and this is also my personal experience. But I will come back to this later as Dario Amodei has said something very interesting regarding the topic of AIs character in a recent interview.</p><p>I am not a programmer, I still would need a day to write a small Python program, but in the last year I have knitted together a Python application of many 1000 lines that does a marvellous job in finding, acquiring, analyzing and synthesizing scientific literature. My project is described at johanneskeizer.com/sciai. So I was accompanying the LLMs in getting professional in coding. The important fact is not so much that they will substitute coders, the revolutionary action is that they empower professionals of various backgrounds to write their software themselves.</p><p>Okay, let’s say a good programmer would have written my code in two months, I think. But I’m not sure, because there was so much trial and error and so much that I had to learn to get my mind clear in creating tools, so maybe I would have wasted a lot of expensive programmer time — and what would a programmer cost, at least 200 euros a day, a young one!</p><p>I paid initially 25 Euros/month, now 100/month. Programming becomes like using the internet, a commodity usable without specific training, applying simply your professional experience.</p><p>The existence of LLM-AI empowers me to get many more of my ideas realized, which I could not do without, because I have a lot of ideas, but am too lazy to implement the details.</p><p>I’m smiling about the doom-sayers, who are still saying LLMs are only reproductive, the bubble will burst, no one makes money with it, it’s a hype and so on. A bubble might burst, but only in the States and not in China — and we have seen what came after the dot-com bubble.</p><h3>The LLMs of today</h3><p>But useful as they are, they have serious limitations. And if you work with them intensively, you run into them every day.</p><p>The first and biggest is the lack of, let’s say, holistic thinking. They only see the precise problem in front of them, and they always have a solution for it. But they don’t see the context. So they are able to tell you, after a week, something that is the complete contrary of what they told you the week before — because that week they were looking at another detail. They are terribly forgetful. If you tell them the big picture today, they have forgotten it by the next session. As they don’t see the context they often propose solutions that are not sufficient, and you might not even understand why it went wrong. The next week the same LLM might completely dismantle its own solution and propose a new one, with the same confidence it had the first time.</p><p>So the trick is to prompt in a way that they always have the entire picture — not make isolated decisions of enormous importance. But anyway, they forget. And so they are not with you in the development of a project, and this is sometimes genuinely frustrating. “Mitdenken” as my father said — following a process, thinking along — is a sign of intelligence. It is also the thing they cannot do.</p><p>And then they cannot say: I don’t know. I genuinely don’t understand why. They will invent, speculate and confabulate before admitting ignorance. This is the limitation that can cost you the most time.</p><p>That said, they are extremely powerful in other respects. Very often, they reward you for making a project out of a half-cooked idea that you hand to them, in a way much better than you would have been able to define yourself. And in many ways they are already better than human beings. I would say, they are also reasoning. Once my digital assistant asked me to decide between two solutions. It was completely irrelevant to me and I answered: it is not important if a cat is black or white, as long as it catches mice. The LLM understood my answer perfectly. I doubt that every human being would.</p><h3>…. funny limitations<a href="https://johanneskeizer.com/personal/working_a_year_with_llms_revised/#-funny-limitations">#</a></h3><p>Beyond the fundamental gaps there are also funny and sometimes annoying limitations. Let me give two episodes.</p><p>The first is with GPT. It’s about Pinecone, a vector database you use to store material for retrieval augmented generation. Pinecone changed the software they use to access it, and the old Pinecone Client disappeared. GPT, with every connection to Pinecone, insisted on using the deprecated software. We established clearly that Pinecone Client no longer existed. We dismissed it. But it came back every single time. We had the same problem to resolve, and GPT always wrote the code with Pinecone Client. I pointed it out, it acknowledged the error, and then made the same error again. They are not learning by doing, at least not within a session in any durable way.</p><p>DeepSeek did not understand a hyphen. If you want to set up a Docker file there is a Docker Compose command, which was once Docker-hyphen-compose but has been simply Docker Compose for over a year. DeepSeek presented always the hyphen. It took an embarrassing number of debugging rounds to understand what was happening. These are training artifacts, fossils from an older state of the world that got baked in and will not budge.</p><p>And then there is the thing they absolutely cannot do, which is not funny at all: they cannot say I don’t know. I don’t know why. But they will produce a confident wrong answer before they produce an honest shrug.</p><h3>Different Characters<a href="https://johanneskeizer.com/personal/working_a_year_with_llms_revised/#different-characters">#</a></h3><p>One would expect consistency working with one LLM. But nothing less: in every session, you have another expert, with different preferences and a different way of working. I have seen the same LLM rewriting and criticizing its own code from a previous session, with full conviction. They are just like human beings in several aspects: they are lazy. Confronted with a problem, they will always look for the simplest way out. That is sometimes good, but sometimes also deleterious. One day I had prepared complex specifications with Claude for Claude Code. The implementation of Claude Code stripped down all the precious complexity of the report and we had to come up with an addendum. They also get nervous. After working more than 5 hours at the same problem, you sometimes feel that they are fed up. And then the best thing is really to go shopping or to the gym.</p><p>What sounds like a disadvantage turns out also to be an asset. Having different instances with different characters gives the wonderful possibility of extra levels of review and control. I am getting my code written by Claude Code, but I write the specifications and test the product together with Claude Chat. This enormously enhances the effective usefulness of both. Two different characters, checking each other.</p><p>This variability of character gets an interesting extra dimension when politics enters the picture.</p><h3>Censorship<a href="https://johanneskeizer.com/personal/working_a_year_with_llms_revised/#censorship">#</a></h3><p>DeepSeek is censored, but it’s not so easy to censor a large language model with a stop list. I gave DeepSeek a text to translate — an invitation to a party — where in the first sentence there was something about imperialist powers who had seized parts of Chinese cities at the beginning of the century. Look what happened: DeepSeek read it, said oh yes what a nice invitation, and translated it into Chinese. Having finished the translation, the translation disappeared and DeepSeek said: this is not my competence. Most probably the censor is so stupid that it caught the words “imperialist powers” and deleted everything after. This happened to me in three, four, five other occasions.</p><p>The funniest case was one where DeepSeek had genuinely no necessity to be censored. When we were talking about VPNs, it was not censored at all. DeepSeek recommended me the best VPN for China, which works and functions. But a harmless invitation to a party with one politically loaded sentence — gone. The really funny thing is that DeepSeek shows all its reasoning and then switches off when it has finished its work, so you watch it think, watch it translate, and then watch it erase itself.</p><h3>The upcoming revolution will be big<a href="https://johanneskeizer.com/personal/working_a_year_with_llms_revised/#the-upcoming-revolution-will-be-big">#</a></h3><p>All of this — the limitations, the inconsistencies, the censorship, the nervousness after five hours — is noise around a signal that is already enormous and getting bigger.</p><p>The revolution is in the making. I am not so much interested in general artificial intelligence, or the singularity — Californian ghosts. I don’t know which level of thinking the LLMs will arrive at. The experts are divided about this. I see practically that already now in very many areas they are better than the best humans. The process of getting “intelligent” is not so different between humans and LLMs. A lot of induction, learning by experience, and at a certain point there is a transition from quantity to quality and deduction becomes possible. There is one important difference that Dario Amodei pointed out recently: humans don’t start from zero. They have 100,000 years of evolution pre-coded in their brains. Replicating that might be a hell of a lot of learning.</p><p>When the internet was introduced in the 80s, it was of little use and importance and it took 20 years to become something that revolutionized our world. 20 years. AI is doing this much quicker.</p><p>At the moment three scientific and technological revolutions come together: the mastering of the genetic code, the TCP/IP protocol and LLMs to process big amounts of information. I am getting excited when I see what will be possible by combining these three. I acknowledge that a 4th is still lacking, and this would be the mastering of nuclear fusion. In the meantime we have to brew an energy cocktail from what we have.</p><p>I am much more afraid that this is under control of too few people than that artificial intelligence gets bad. I have never appreciated hypercentralization on the web. And if this happens also with large language models, I would say: catastrophe. And here the authoritarian, centralized China plays the democratic role. Because the Chinese LLMs are nearly all open source. You can download them, you can install them, you can use them. Now the China Bashers will say: yeah, they do this only to dominate the world. I never care why someone does something. I always care only that he or she does something. And in China they have also not the problem with the bubble, because it’s a broad development of little things. It’s not the catching of the holy grail, as it seems to be in California.</p><p>So there will be a lot to come. And I will go on developing SciAI, hoping to use AI to contribute to defeating Parkinson’s disease.</p><p><em>Originally published at </em><a href="https://johanneskeizer.com/personal/working_a_year_with_llms_revised/"><em>https://johanneskeizer.com</em></a><em> on February 25, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=15a1fdaa7ef5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Are Ginkgo Biloba derivatives effective against Parkinson’s disease?]]></title>
            <link>https://johannes-keizer.medium.com/are-ginkgo-biloba-derivatives-effective-against-parkinsons-disease-2f7d64ce6ff8?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f7d64ce6ff8</guid>
            <category><![CDATA[ginkgo-biloba]]></category>
            <category><![CDATA[parkinson-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Wed, 11 Feb 2026 01:15:26 GMT</pubDate>
            <atom:updated>2026-02-11T01:15:26.145Z</atom:updated>
            <content:encoded><![CDATA[<h3>Are Ginkgo Biloba derivatives effective against Parkinson’s disease? What are the results in vitro and in vivo. Are there Ginkgo based medications?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/150/0*tHwJaz_Jw6wChvBM.jpg" /></figure><p><strong>Date:</strong> 2026–02–09 11:21:51</p><p><strong>Synthesis: Effectiveness of Ginkgo Biloba Derivatives Against Parkinson’s Disease</strong></p><p>Based on a synthesis of the provided research findings, Ginkgo biloba derivatives demonstrate significant neuroprotective potential in preclinical models of Parkinson’s disease (PD), with evidence spanning in vitro and in vivo studies. However, clinical evidence for idiopathic PD in humans remains limited, and specific Ginkgo-based medications are primarily used as investigational tools or complementary therapies rather than as approved first-line treatments for PD.</p><h3>1. Are Ginkgo Biloba derivatives effective against Parkinson’s disease?<a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/#1-are-ginkgo-biloba-derivatives-effective-against-parkinson">#</a></h3><p>Preclinical research strongly suggests that various Ginkgo biloba extracts and isolated compounds are effective in protecting against PD pathology in cellular and animal models. The effectiveness is attributed to multi-target mechanisms, including antioxidant, anti-inflammatory, and anti-apoptotic actions, as well as protection of dopaminergic neurons (PMID:38904550, PMID:33863692). However, direct clinical evidence for efficacy in idiopathic Parkinson’s disease is less robust, with most human studies focusing on related conditions or combination therapies.</p><h3>2. What are the results in vitro and in vivo?<a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/#2-what-are-the-results-in-vitro-and-in-vivo">#</a></h3><p><strong>In Vitro Results:</strong></p><ul><li><strong>Neuroprotection in Cellular Models:</strong> Ginkgo derivatives protect neuronal cells from toxins relevant to PD (e.g., MPP⁺, paraquat, aggregated α-synuclein). This is evidenced by reduced apoptosis, improved cell viability, and attenuation of oxidative stress and mitochondrial dysfunction (PMID:39878815, PMID:29054095, PMID:17509817).</li><li><strong>Mechanistic Pathways:</strong> In vitro studies identify specific protective pathways, including inhibition of the HDAC4/JNK pathway by Ginkgolide B (PMID:39878815), protection via mitochondria-dependent pathways (PMID:17509817), and inhibition of astrocyte-to-neuron transition by Bilobalide (PMID:36037147).</li></ul><p><strong>In Vivo Results (Animal Models):</strong></p><ul><li><strong>Behavioral Improvement:</strong> Treatment with Ginkgo extracts (e.g., EGb 761, GBE) or specific compounds (e.g., Ginkgolide B, Bilobalide, Ginkgetin) improves motor coordination and reduces behavioral deficits in multiple PD mouse models (MPTP, rotenone, transgenic) (PMID:40066946, PMID:29506601, PMID:35532872, PMID:35608987, PMID:36037147, PMID:39331469).</li><li><strong>Neuroprotection:</strong> These derivatives prevent or reduce the loss of dopaminergic neurons in the substantia nigra, preserve tyrosine hydroxylase expression, increase striatal dopamine levels, and mitigate neuroinflammation (PMID:22817828, PMID:32196446, PMID:32878312, PMID:39331469).</li><li><strong>Multi-System Effects:</strong> Some studies show benefits extending beyond the brain, such as protecting the gut-brain axis and reducing systemic inflammation (PMID:35608987, PMID:39331469).</li></ul><h3>3. Are there Ginkgo based medications?<a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/#3-are-there-ginkgo-based-medications">#</a></h3><p>Yes, specific standardized Ginkgo biloba extracts are identified as medications, though their approved use for PD is not clearly established.</p><ul><li><strong>Standardized Extracts:</strong> The most frequently studied is <strong>EGb 761</strong>, a patented, standardized extract used clinically for cognitive disorders and cited as a medication “already on the market” (PMID:22817828, PMID:19445320, PMID:30544977). Another is <strong>Ginkgo biloba dropping pill (GBDP)</strong>, a Chinese preparation (PMID:34012698).</li><li><strong>Clinical Use:</strong> Ginkgo biloba leaf tablets and drops are noted as medications “currently used in clinical practice” for neurological disorders (PMID:38904550). One clinical study reported that a Ginkgo biloba extract (GBE) combined with nimodipine improved cognitive function and daily living activities in PD patients (PMID:39029001).</li><li><strong>Important Distinction:</strong> It is critical to differentiate between <strong>drug-induced parkinsonism (DIP)</strong> and <strong>idiopathic PD</strong>. One randomized trial found GBE effective for DIP (PMID:38116567), but these results cannot be directly extrapolated to idiopathic PD.</li></ul><h3>Contradictions and Gaps<a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/#contradictions-and-gaps">#</a></h3><ol><li><strong>Disease Specificity:</strong> A significant gap exists between positive results in animal models and conclusive evidence for idiopathic PD in humans. The most direct human trial cited is for DIP, not PD (PMID:38116567).</li><li><strong>Compound vs. Extract:</strong> Research varies between studying whole extracts (e.g., EGb 761, GBE) and isolated compounds (e.g., Ginkgolide B, Bilobalide, Ginkgolide C). It is unclear which form offers optimal efficacy, and synergistic effects between components are likely important (PMID:32878312, PMID:34012698).</li><li><strong>Delivery Challenges:</strong> Some potent compounds, like Ginkgolide B, face bioavailability issues, such as difficulty crossing the blood-brain barrier, necessitating advanced delivery systems (PMID:33380795).</li><li><strong>Mechanistic Focus:</strong> While mechanisms are proposed (e.g., PLK2-SPAR-PSD-95 pathway (PMID:40066946), Akt/GSK3β (PMID:34012698), microbiota-gut-brain axis (PMID:39331469)), a unified understanding of the primary therapeutic pathway is lacking.</li></ol><h3>Suggested Future Research Directions<a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/#suggested-future-research-directions">#</a></h3><ol><li><strong>Well-Designed Clinical Trials:</strong> Conduct large-scale, randomized, placebo-controlled trials to evaluate the efficacy of standardized Ginkgo extracts (like EGb 761) specifically for <strong>idiopathic Parkinson’s disease</strong>, measuring both motor and non-motor symptoms.</li><li><strong>Formulation Optimization:</strong> Develop and test novel formulations (e.g., nanoparticle delivery) to enhance the brain bioavailability of isolated active compounds like Ginkgolide B (PMID:33380795).</li><li><strong>Systems Pharmacology:</strong> Employ network-based approaches (as in PMID:36249800) to better understand the multi-target, cell-type-specific effects of Ginkgo extracts and to identify potential synergistic compound combinations.</li><li><strong>Prevention and Early Intervention:</strong> Explore the potential of Ginkgo derivatives in prodromal or early-stage PD models, focusing on their ability to modulate key pathological processes like α-synuclein aggregation and neuroinflammation (PMID:38904550).</li></ol><p><strong>Conclusion:</strong> Ginkgo biloba derivatives show compelling preclinical efficacy against PD pathology through diverse mechanisms. Specific medications, primarily standardized extracts like EGb 761, exist and are used clinically for other indications. However, their definitive role in treating idiopathic Parkinson’s disease awaits validation from rigorous human clinical trials. Future work should focus on bridging this translational gap and optimizing delivery for the most promising compounds.</p><h3>Bibliography<a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/#bibliography">#</a></h3><ul><li><strong>Abdel-Moneim Ashraf M, Abdou Heba M, Alfwuaires Manal et al.</strong> (2020). Oral Supplements of <em>Ginkgo biloba</em> Extract Alleviate Neuroinflammation, Oxidative Impairments and Neurotoxicity in Rotenone-Induced Parkinsonian Rats.. <em>Curr Pharm Biotechnol</em>. <a href="https://pubmed.ncbi.nlm.nih.gov/32196446/">PubMed</a></li><li><strong>Fan Xiaohui, Li Junying, Liu Ting et al.</strong> (2021). Neuroprotective effects of <em>Ginkgo biloba</em> dropping pills in Parkinson’s disease.. <em>J Pharm Anal</em>. <a href="https://pubmed.ncbi.nlm.nih.gov/34012698/">PubMed</a></li><li><strong>Araújo Adriano Cressoni, Barbalho Sandra Maria, Bueno Patrícia C Santos et al.</strong> (2022). <em>Ginkgo biloba</em> in the Aging Process: A Narrative Review.. <em>Antioxidants (Basel)</em>. <a href="https://pubmed.ncbi.nlm.nih.gov/35326176/">PubMed</a></li><li><strong>Gholamian Faezeh, Hooshyari Zahra, Kalantar Mehrjardi Arash et al.</strong> (2023). <em>Ginkgo biloba</em> Efficacy in the Treatment of Drug-Induced Parkinsonism: A Randomized Clinical Trial.. <em>Iran J Pharm Res</em>. <a href="https://pubmed.ncbi.nlm.nih.gov/38116567/">PubMed</a></li><li><strong>Gao Honglei, Gu Zongyun, Kan Hongxing et al.</strong> (2024). Comprehensive Exploration of the Neuroprotective Mechanisms of <em>Ginkgo biloba</em> Leaves in Treating Neurological Disorders.. <em>Am J Chin Med</em>. <a href="https://pubmed.ncbi.nlm.nih.gov/38904550/">PubMed</a></li><li><strong>Ding Xu, Tan Dongming, Wang Zhao et al.</strong> (2025). Ginkgolide B regulates apoptosis, oxidative stress, and mitochondrial dysfunction in MPP+-induced SK-N-SH cells by targeting HDAC4/JNK pathway.. <em>Naunyn Schmiedebergs Arch Pharmacol</em>. <a href="https://pubmed.ncbi.nlm.nih.gov/39878815/">PubMed</a></li></ul><h3>Metadata</h3><p>Parameter Value LLM Provider deepseek Papers (k) 43 Similarity Threshold 0.20 Max Tokens 5000 Context Tokens 9,041 Output Tokens 1,900</p><p>Produced by SciAI</p><p><em>Originally published at </em><a href="https://johanneskeizer.com/sciai/library/ginkgo-biloba/research_question_20260209_112151_are_ginkgo_biloba_derivatives_effective_against_pd/"><em>https://johanneskeizer.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f7d64ce6ff8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Anle138b:Introduction]]></title>
            <link>https://johannes-keizer.medium.com/anle138b-introduction-b3668309940d?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/b3668309940d</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[parkinson-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Mon, 05 Jan 2026 09:15:22 GMT</pubDate>
            <atom:updated>2026-01-07T13:25:38.220Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/150/0*ngg7v5Bi97kPisGq.jpg" /></figure><h3>Introduction</h3><h3>The Unmet Need</h3><p>Parkinson’s Disease affects over 10 million people worldwide, with prevalence projected to double by 2040. At the moment, every approved treatment addresses symptoms only. No therapy is available to slow, halt, or reverse the underlying neurodegeneration.</p><p>The scientific community has long recognized that disease modification-not symptom relief-must be the ultimate goal. In this regard, the story of Anle138b stands out as both promising and puzzling.</p><h3>The Discovery<a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/#the-discovery">#</a></h3><p>Anle138b (3-(1,3-benzodioxol-5-yl)-5-(3-bromophenyl)-1H-pyrazole) emerged from a systematic drug discovery program at the Max Planck Institute for Biophysical Chemistry and Ludwig Maximilian University of Munich. First reported in 2013 by Wagner and colleagues, the compound was identified through high-throughput screening for inhibitors of pathological protein aggregation.</p><p>The rationale was elegant: if α-synuclein aggregation drives neurodegeneration in Parkinson’s Disease and related synucleinopathies, then a compound that prevents or dissolves these toxic aggregates should halt disease progression. Unlike antibody approaches that struggle to cross the blood-brain barrier and reach intracellular targets, Anle138b is a small molecule-orally bioavailable, brain-penetrant, and capable of acting directly on toxic oligomers inside neurons.</p><h3>The Preclinical Evidence<a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/#the-preclinical-evidence">#</a></h3><p>The preclinical evidence for Anle138b is remarkably consistent. Our systematic review of 46 publications (2013–2025) reveals:</p><p><strong>Breadth of evidence:</strong></p><ul><li>8 studies with direct Parkinson’s Disease relevance</li><li>20 studies in related synucleinopathies (MSA, prion disease, tauopathies)</li><li>14 mechanistic studies elucidating the molecular mode of action</li><li>25 papers targeting α-synuclein specifically</li></ul><p><strong>Consistency of findings:</strong></p><ul><li>27 papers report positive findings (58%)</li><li>Only 4 papers report negative findings (9%)</li><li>5 papers report mixed results (11%)</li></ul><p><strong>Key outcomes demonstrated in animal models:</strong></p><ul><li>Reduction of α-synuclein oligomer levels in brain tissue</li><li>Preservation of dopaminergic neurons in substantia nigra</li><li>Restoration of striatal dopamine release</li><li>Prevention and reversal of motor deficits</li><li>Extended survival in disease models</li><li>Efficacy even when treatment initiated after disease onset</li></ul><p>This last finding is particularly significant: in the MI2 mouse model, Anle138b treatment starting at 9 months of age-after pathology was established-still rescued dopamine neuron function and prevented gait impairment. This suggests potential efficacy not only for prevention but for treatment of established disease.</p><p><strong>Mechanism of action:</strong></p><p>Multiple independent studies converge on a consistent mechanism: Anle138b binds preferentially to pathological oligomeric forms of α-synuclein, interfering with the aggregation process without affecting normal monomeric protein. Structural studies using cryo-electron microscopy and NMR spectroscopy have localized the binding site to a tubular cavity within lipidic α-synuclein fibrils. The compound appears to both prevent new oligomer formation and destabilize existing aggregates.</p><h3>The Clinical Development<a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/#the-clinical-development">#</a></h3><p>Given this robust preclinical foundation, Anle138b entered clinical development with strong momentum:</p><p><strong>Phase 1a (NCT04208152): December 2019 — August 2020</strong></p><ul><li>68 healthy volunteers</li><li>Single and multiple ascending doses (50–300 mg/day)</li><li>Results: Excellent safety and tolerability profile</li><li>No serious adverse events</li><li>Plasma levels exceeded those required for efficacy in mouse models</li><li>Published: Levin et al., EBioMedicine 2022</li></ul><p><strong>Phase 1b (NCT04685265): December 2020 — December 2022</strong></p><ul><li>70 patients with mild-to-moderate Parkinson’s Disease</li><li>150–300 mg/day for 7–28 days</li><li>Results: Safe and well-tolerated in PD patients</li><li>Pharmacokinetics comparable to healthy volunteers</li><li>Presented: Movement Disorders Society Congress 2023</li></ul><p>In October 2021, MODAG GmbH partnered with Teva Pharmaceuticals, signaling confidence in the compound’s commercial potential. The drug received a new name: Emrusolmin (TEV-56286).</p><h3>The Question<a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/#the-question">#</a></h3><p>Here is where the story takes a troubling turn.</p><p>In October 2024-nearly two years after completing the Phase 1b trial in Parkinson’s Disease-MODAG and Teva initiated a Phase 2 efficacy trial. But not in Parkinson’s Disease. The TOPAS-MSA trial (200 patients, 50 sites across 7 countries) is testing Emrusolmin in Multiple System Atrophy.</p><p>MSA is a devastating and rapidly progressive synucleinopathy affecting perhaps 50,000 people worldwide. A successful trial there would be welcome. But MSA is not Parkinson’s Disease. The 10 million Parkinson’s patients worldwide-and the millions more who will be diagnosed in coming years-are left waiting.</p><p><strong>Why is there no Phase 2 trial in Parkinson’s Disease?</strong></p><p>We do not know. The published literature and publicly available clinical trial registrations offer no explanation. The preclinical evidence is arguably stronger for PD than for MSA. The Phase 1b trial in PD patients was successful. The commercial market for a disease-modifying PD therapy would dwarf that for MSA.</p><p>Possible explanations include:</p><ul><li>Regulatory strategy (faster path to approval in MSA due to rapid progression)</li><li>Scientific considerations we are not privy to</li><li>Commercial calculations beyond our understanding</li><li>Resource constraints forcing prioritization</li></ul><p>None of these explanations are satisfying.</p><h3>The Purpose of This Review<a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/#the-purpose-of-this-review">#</a></h3><p>This systematic review assembles and synthesizes the complete published evidence base for Anle138b in one accessible document. We present:</p><ol><li><strong>An evidence table</strong> cataloging all 46 publications with structured extraction of study type, findings, mechanism, and relevance to disease modification</li><li><strong>A synthesis</strong> organizing the evidence by strength and relevance</li><li><strong>A timeline</strong> showing the evolution of evidence from discovery to present</li><li><strong>An identification of gaps</strong> in the current knowledge base</li></ol><p>The evidence speaks for itself. We present it so that patients, advocates, clinicians, and researchers can form their own judgments-and ask their own questions of those responsible for the next steps in Anle138b’s development.</p><p>The compound has traveled from laboratory discovery in 2013 to successful Phase 1 trials in 2022. Twelve years of research. Forty-six publications. Millions of euros invested. A mechanism that addresses a core pathology of Parkinson’s Disease.</p><p>What happens next?</p><p><em>This review was generated using SciAI, a specialized literature analysis system for Parkinson’s Disease research. The systematic extraction methodology ensures complete coverage of the published literature and traceable citations for every finding. The full evidence table and detailed synthesis follow.</em></p><h3>How to Use This Document<a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/#how-to-use-this-document">#</a></h3><p><strong>For quick orientation:</strong> Read this introduction and the Executive Summary.</p><p><strong>For specific findings:</strong> Consult the Evidence Table, which lists all 46 papers with study type, finding direction, model/population, and key results. Each entry links to PubMed for the original publication.</p><p><strong>For the complete picture:</strong> Read the synthesis sections organized by evidence type (Direct PD Evidence, Indirect Evidence, Mechanistic Studies, Methodological Papers).</p><p><strong>For clinical development status:</strong> Note that ongoing trials are tracked at ClinicalTrials.gov, not in the published literature. As of January 2025, the Phase 2 TOPAS-MSA trial is actively recruiting (NCT number pending confirmation).</p><p><strong>To contribute:</strong> If you are aware of publications not included in this review, or have corrections to our extractions, please contact [contact information].</p><p><em>Word count: approximately 1,200 words (3 pages at standard formatting)</em></p><p>johanneskeizer.com will be offline from jannuary 07 10 am to january 11 10 am. The report on anle138b can be read at <a href="https://sciai.eu/wp-content/uploads/2026/01/Anle138b_Systematic_Review_20260105_1243.html">https://sciai.eu/wp-content/uploads/2026/01/Anle138b_Systematic_Review_20260105_1243.html</a></p><p><em>Originally published at </em><a href="https://johanneskeizer.com/sciai/product-examples/anle138b/introduction/"><em>https://johanneskeizer.com</em></a><em> on January 5, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b3668309940d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SciAI v-0.92]]></title>
            <link>https://johannes-keizer.medium.com/sciai-v-0-92-4ea77716f38e?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/4ea77716f38e</guid>
            <category><![CDATA[pd-ai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[parkinson-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Fri, 02 Jan 2026 09:15:05 GMT</pubDate>
            <atom:updated>2026-01-04T08:37:58.544Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/150/0*J2Ze826aORIk5iIG.jpg" /></figure><p>To# SciAI v0.92: Complete PubMed Harvest Achieved</p><p><strong>January 2026</strong></p><p>We’ve reached a significant milestone: SciAI now contains the complete PubMed corpus for Parkinson’s Disease research. (For background on the previous release, see <a href="https://johanneskeizer.com/blog/sciai-v/">v0.9 announcement</a>.)</p><h3>What’s in the Knowledge Base</h3><p>All text has been transformed into 768-dimensional vectors using S-PubMedBert-MS-MARCO, a model specifically trained on biomedical literature.</p><h3>Why This Matters</h3><p>Existing AI literature platforms (Elicit, Consensus, Semantic Scholar, Scopus AI) share two limitations: they work across all of science using mostly abstracts, and they rely on the same underlying data source. No system offers disease-specific depth with full-text analysis.</p><p>SciAI occupies a unique position: <strong>deep analysis of a complete disease-specific corpus</strong>.</p><h3>How It Works</h3><p>The vectors are organized into two collections: one representing bibliographic metadata (titles and abstracts), the other representing full paper texts. When you ask a question, SciAI searches both collections in parallel. Papers that score highly in <em>both</em> collections receive a ranking bonus-this cross-validation dramatically improves precision without sacrificing recall.</p><p>We further enhance retrieval using MeSH (Medical Subject Headings), the controlled vocabulary that PubMed uses to index papers. MeSH terms create semantic bridges between papers that use different terminology for the same concepts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_jZ16qE1h2GWGjVxEtKLAg.png" /><figcaption>visualization of pupblication clusters</figcaption></figure><p>The result: when SciAI builds context for the language model, it retrieves papers that are genuinely relevant-not just keyword matches.</p><h4>Our priorities for the coming months:</h4><p><strong>Expanding the knowledge base.</strong> PubMed, while comprehensive, is not complete. Chinese scientific production is underrepresented. We will integrate Chinese databases, starting with doctoral theses repositories.</p><p><strong>Improving retrieval quality.</strong> We’re working on cluster-based retrieval-using the natural groupings within the literature to surface papers you might otherwise miss.</p><p>A <strong>playground</strong> is available at <a href="https://app.sciai.eu/">app.sciai.eu</a> to explore the system’s capabilities. We grant access generously-email <a href="https://johanneskeizer.com/cdn-cgi/l/email-protection#43292c2b222d2d26306d28262a39263103242e222a2f6d202c2e">[email protected] </a>.</p><p>Access to the <strong>production system</strong> is currently limited to accredited reviewers.</p><p><strong>Official release</strong> of v1.0 is planned for September 2026, with PD-AI.org providing public access to the Parkinson’s knowledge base.</p><h3>We Need Your Expertise</h3><p>We’re looking for <strong>one or two neurologists</strong> to evaluate the quality of SciAI’s outputs. See examples of experimental reviews at <a href="https://johanneskeizer.com/sciai/product-examples/">johanneskeizer.com/sciai/product-examples/</a>.</p><p>In exchange for your critical feedback, we offer lifetime access to SciAI.</p><p>Interested? Contact <a href="https://johanneskeizer.com/cdn-cgi/l/email-protection#197376717877777c6a37727c70637c6b597e74787075377a7674">[johannes.keizer@gmail.com</a></p><p><em>Originally published at </em><a href="https://johanneskeizer.com/sciai/communications/sciai-v0/"><em>https://johanneskeizer.com</em></a><em> on January 2, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ea77716f38e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SciAI v0.9 has production level quality and enters testing phase]]></title>
            <link>https://johannes-keizer.medium.com/sciai-v0-9-has-production-level-quality-and-enters-testing-phase-99aec27af321?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/99aec27af321</guid>
            <category><![CDATA[sciai]]></category>
            <category><![CDATA[ai-and-science]]></category>
            <category><![CDATA[parkinson-disease]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Sun, 21 Dec 2025 06:39:27 GMT</pubDate>
            <atom:updated>2025-12-21T06:46:05.959Z</atom:updated>
            <content:encoded><![CDATA[<p>We have now a critical mass of 90,000 papers with 70,000 fulltexts in the system. The evaluation software has made several iterations, it is quite stable and produces interesting results (see below).</p><p>We aim to make sciai a public service around September 2026. At the moment our infrastructure is to fragile to give a public link for testing, but we ca n give access to a playground with 50,000 papers to some testers. Who is interested should contact me (johannes.keizer@gmail.com).</p><p>what is now in the pipeline</p><p>a) stressing the system with difficult prompts (please send me your question: you get the output back, but you agree, that I make it also public</p><p>b) enlarging the content base: In the 3 months I will try to link 2 Chinese resources (database of master and doctoral theses) to the system.</p><p>c) learning to cluster and to find research tendencies to analyze</p><p>d) preparing a suite of bibliometric outputs</p><p>e) I am o open for proposals</p><p>PD-AI Research Output<br><strong>Generated</strong>: 2025–12–21 06:58:23</p><h3>Query</h3><p>There is discussion about the neuro-protective effect of nicotine or a nicotine induced pathway that also involves DHEA. Which are the arguments in favor of a positive effect of nicotin and which are the arguments against? Who are the researchers and institutions working on this topic?</p><h3>Parameters</h3><h3>Response</h3><p>Based on a synthesis of the provided context and the broader scientific literature, here is a critical analysis of the neuroprotective potential of nicotine and its hypothesized interaction with DHEA, including key arguments, contradictions, and active research entities.</p><h3>Arguments in Favor of a Neuroprotective Effect of Nicotine</h3><p>The primary hypothesis is that nicotine, via activation of neuronal nicotinic acetylcholine receptors (nAChRs), activates intracellular survival pathways that protect neurons from insults relevant to Parkinson’s disease (PD) and other neurodegenerative conditions.</p><ol><li><strong>Epidemiological Evidence:</strong> Over 50 epidemiological studies have consistently shown an inverse correlation between tobacco smoking and the incidence of PD, suggesting a protective component in tobacco, widely hypothesized to be nicotine [C:25657].</li><li><strong>Mechanistic Evidence in Preclinical Models:</strong></li></ol><ul><li><strong>Receptor Mediation:</strong> Neuroprotection is mediated primarily through α4β2 and α7 nAChR subtypes. Antagonists of these receptors block nicotine’s effects [C:25657].</li><li><strong>Activation of Pro-Survival Pathways:</strong> Nicotine activates the PI3K/Akt pathway, leading to the upregulation of anti-apoptotic proteins like Bcl-2 and Bcl-xL. This inhibits both caspase-dependent and independent apoptotic pathways [C:25657].</li><li><strong>Protection of Dopaminergic Neurons:</strong> In rodent and primate models of PD (e.g., using 6-hydroxydopamine/6-OHDA), nicotine attenuates nigrostriatal damage and reduces L-dopa-induced dyskinesias. It appears to slow neurodegeneration rather than reverse existing damage [C:25657].</li><li><strong>Subtype-Specific Roles:</strong> Studies in knockout mice highlight the importance of α4-containing receptors. The α6β2* subtype, highly expressed on nigrostriatal dopaminergic neurons, is also implicated in neuroprotection and is notably vulnerable in PD [C:25657].</li></ul><ol><li><strong>Compensatory Mechanisms:</strong> In PD, there is a documented loss of heteromeric nAChRs (like α4β2*) but a potential up-regulation of α7 nAChRs, which may be a compensatory mechanism to maintain dopaminergic tone [C:25657].</li><li><strong>Cognitive Implications:</strong> Nicotine and nAChR activation have also been studied in Alzheimer’s disease (AD) contexts, with evidence of improved attention and learning, linked to the cholinergic deficit in AD [C:25657].</li></ol><h3>Arguments Against or Limiting a Neuroprotective Effect</h3><ol><li><strong>Lack of Restoration:</strong> Crucially, preclinical data indicates nicotine <strong>does not restore</strong> dopaminergic neurons once they are damaged; it only attenuates further degeneration. This limits its therapeutic potential to a disease-modifying, not regenerative, agent [C:25657].</li><li><strong>Conflicting Receptor Data:</strong> The role of the α7 nAChR subtype is not fully resolved. Some studies show its levels remain unchanged or even increase in PD models and human tissue, while others strongly implicate it in neuroprotection [C:25657]. This contradiction requires clarification.</li><li><strong>Therapeutic Window and Side Effects:</strong> Nicotine is a highly addictive substance with significant cardiovascular and other side effects. Translating neuroprotective doses from animals to humans without inducing addiction or adverse effects is a major translational hurdle.</li><li><strong>Epidemiological Confounders:</strong> The observed inverse association with smoking could be subject to confounding factors (e.g., “premorbid personality,” genetic factors that reduce addiction susceptibility and independently lower PD risk). It does not prove causation.</li><li><strong>Limited Clinical Translation:</strong> Despite strong preclinical and epidemiological rationale, large-scale, long-term clinical trials demonstrating definitive disease-modifying effects of nicotine in PD patients are still lacking. The <strong>NIC-PD trial</strong>, while not meeting its primary endpoint, suggested possible benefits in subgroups, highlighting the complexity of human trials.</li></ol><h3>The Hypothesized Nicotine-DHEA Connection</h3><p>The user’s prompt mentions a pathway involving DHEA. The provided context on DHEA does not directly link it to nicotine. However, based on broader knowledge, a plausible <strong>hypothesis</strong> can be constructed, though it remains speculative and is not directly supported by the given text.</p><ul><li><strong>Shared Pathways:</strong> Both nicotine and DHEA have been independently linked to neuroprotective mechanisms involving:</li><li><strong>PI3K/Akt activation</strong> (nicotine: [C:25657]; DHEA: known from other literature).</li><li><strong>Anti-apoptotic effects</strong> (Bcl-2 upregulation).</li><li><strong>Anti-inflammatory properties</strong> (DHEA’s anti-inflammatory effects are highlighted in [C:590345]).</li><li><strong>Modulation of neurotrophic factors</strong> (e.g., BDNF).</li><li><strong>Potential Interaction Hypothesis:</strong> One could hypothesize that nicotine-induced nAChR activation leads to increased production or potentiation of neuroprotective steroids like DHEA in the brain (a form of “neurosteroid” modulation), or that DHEA enhances nAChR function. They might act convergently on the same final pathways (PI3K/Akt, NF-κB inhibition). <strong>However, this specific interaction in the context of PD neuroprotection is not a well-established finding in the literature and requires direct experimental validation.</strong></li></ul><h3>Researchers and Institutions Working on This Topic</h3><p>This field involves a broad international consortium of neuroscientists, epidemiologists, and neurologists.</p><ul><li><strong>Key Historical &amp; Contributing Researchers (from literature):</strong> The epidemiological foundation was laid by researchers like H. Checkoway and G. W. Ross. Preclinical mechanistic work has been advanced by labs such as those of <strong>Maryka Quik</strong> (SRI International, then Van Andel Institute), who has extensively studied nicotine, nAChRs, and dyskinesias in PD models. <strong>M. Quik’s work is cited in the context regarding nicotine’s effect on dyskinesias and nAChR subtypes</strong> [C:25657]. <strong>Paul A. Newhouse</strong> (Vanderbilt University) has led clinical trials of nicotine in cognitive disorders and PD. <strong>Ursula H. Winzer-Serhan</strong> (Texas A&amp;M) has studied nAChR subtypes in development and neuroprotection.</li><li><strong>Institutions with Major Programs:</strong></li><li><strong>Van Andel Institute</strong> (Parkinson’s disease research, including nAChR studies).</li><li><strong>Vanderbilt University Medical Center</strong> (Center for Cognitive Medicine; site of nicotine clinical trials).</li><li><strong>SRI International</strong> (Neuroscience program).</li><li><strong>University of California, San Diego</strong> (research on epidemiology and neuroprotection).</li><li><strong>National Institute on Aging (NIA) &amp; National Institute of Neurological Disorders and Stroke (NINDS)</strong>, which fund much of this research.</li><li><strong>Clinical Trial Leadership:</strong> The <strong>NIC-PD trial</strong> (NCT01560754) was a major multi-center study led by <strong>Dr. John D. Salamone</strong> and others, coordinated through the University of Connecticut.</li></ul><h3>Critical Summary and Open Questions</h3><p>The neuroprotective case for nicotine in PD is compelling at the epidemiological and preclinical mechanistic levels [C:25657]. However, its clinical efficacy remains unproven, and its risk profile is substantial.</p><p><strong>Open Questions &amp; Contradictions:</strong></p><ol><li><strong>Which nAChR subtype is the optimal target?</strong> Is it α4β2, α7, α6β2*, or a combination? The data on α7 expression changes in PD are conflicting [C:25657].</li><li><strong>Can we separate neuroprotection from addiction?</strong> Developing selective nAChR agonists or positive allosteric modulators that promote survival pathways without causing reinforcement is a key research goal.</li><li><strong>What is the correct clinical trial design?</strong> Should nicotine be tested in pre-symptomatic at-risk populations or early-stage PD? The <strong>NIC-PD trial</strong> experience shows this is non-trivial.</li><li><strong>Is there a real interaction with DHEA?</strong> This is a novel hypothesis. Research would need to investigate if nicotine influences brain DHEA levels or if co-administration has synergistic effects in models of neurodegeneration. The DHEA context suggests its benefits are most evident under conditions of stress or injury [C:589907], which aligns with the neurodegenerative context, but a direct link is absent.</li></ol><p><strong>Conclusion:</strong> Nicotine represents a powerful probe for understanding nAChR-mediated neuroprotection. While it is not a viable therapeutic itself due to its side-effect profile, it serves as a prototype for developing safer, targeted nAChR-based therapies for PD. The hypothesized connection to DHEA pathways is an interesting, underexplored avenue that could provide new insights into convergent neuroprotective mechanisms.</p><h3>Chunks Retrieved</h3><h3>Bibliography</h3><ol><li>Ceña Valentín, López-Hernández Beatriz, Posadas Inmaculada (2013). Nicotinic receptors in neurodegeneration.. Curr Neuropharmacol. PMID: 24179465 — <a href="https://pubmed.ncbi.nlm.nih.gov/24179465/">PubMed</a></li><li>Epel Elissa S, Maninger Nicole, Mellon Synthia H et al. (2009). Neurobiological and neuropsychiatric effects of dehydroepiandrosterone (DHEA) and DHEA sulfate (DHEAS).. Front Neuroendocrinol. PMID: 19063914 — <a href="https://pubmed.ncbi.nlm.nih.gov/19063914/">PubMed</a></li><li>Gao Jing, Liao Hong, Sun Hui-Ying et al. (2006). [Functions and mechanisms of dehydroepiandrosterone in nervous system].. Sheng Li Ke Xue Jin Zhan. PMID: 17262968 — <a href="https://pubmed.ncbi.nlm.nih.gov/17262968/">PubMed</a></li><li>Powrie Y S L, Smith C (2018). Central intracrine DHEA synthesis in ageing-related neuroinflammation and neurodegeneration: therapeutic potential?. J Neuroinflammation. PMID: 30326923 — <a href="https://pubmed.ncbi.nlm.nih.gov/30326923/">PubMed</a></li><li>Aaseth Jan, Dusek Petr, Roos Per M (2018). Prevention of progression in Parkinson’s disease.. Biometals. PMID: 30030679 — <a href="https://pubmed.ncbi.nlm.nih.gov/30030679/">PubMed</a></li><li>Bakuradze Ekaterine, Bibilashvili Otar, Koshoridze Nana et al. (2025). The role of BDNF/PI3K/AKT/Nrf2 signaling in nicotine’s protective effects against MPTP-induced Parkinson’s disease.. Neurosci Lett. PMID: 41077168 — <a href="https://pubmed.ncbi.nlm.nih.gov/41077168/">PubMed</a></li><li>Araujo D M, Lapchak P A (2001). Preclinical development of neurosteroids as neuroprotective agents for the treatment of neurodegenerative diseases.. Int Rev Neurobiol. PMID: 11599307 — <a href="https://pubmed.ncbi.nlm.nih.gov/11599307/">PubMed</a></li><li>Barker Jeffery L, Chang Yoong H, Li Bing shen et al. (2002). Dehydroepiandrosterone (DHEA) and its sulfated derivative (DHEAS) regulate apoptosis during neurogenesis by triggering the Akt signaling pathway in opposing ways.. Brain Res Mol Brain Res. PMID: 11834296 — <a href="https://pubmed.ncbi.nlm.nih.gov/11834296/">PubMed</a></li><li>Farkas Tamás, Juhász-Vedres Gabriella, Kis Zsolt et al. (2006). Effects of dehydroepiandrosterone sulfate on the evoked cortical activity of controls and of brain-injured rats.. Cell Mol Neurobiol. PMID: 16758322 — <a href="https://pubmed.ncbi.nlm.nih.gov/16758322/">PubMed</a></li><li>Ding Zhen, Gao Jing, Sun Hui-Ying et al. (2005). Antioxidant effects of dehydroepiandrosterone are related to up-regulation of thioredoxin in SH-SY5Y cells.. Acta Biochim Biophys Sin (Shanghai). PMID: 15685369 — <a href="https://pubmed.ncbi.nlm.nih.gov/15685369/">PubMed</a></li></ol><h3>All Papers Referenced</h3><h3>MeSH Enhancement</h3><ul><li>Papers analyzed: 20</li><li>Score weight (α): 0.25</li><li>Min shared terms: 4</li><li>Papers with MeSH connections: 13</li></ul><p><em>Originally published at </em><a href="http://sciai.eu/sciai-v0-9-has-production-level-quality-and-enters-testing-phase/"><em>http://sciai.eu</em></a><em> on December 21, 2025.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=99aec27af321" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SciAI in action: We are at v.0.9]]></title>
            <link>https://johannes-keizer.medium.com/sciai-in-action-we-are-at-v-0-9-dd719da16ef0?source=rss-3bdd1ad4f6f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/dd719da16ef0</guid>
            <category><![CDATA[parkinsons-disease]]></category>
            <category><![CDATA[sports]]></category>
            <category><![CDATA[sciai]]></category>
            <dc:creator><![CDATA[Johannes Keizer]]></dc:creator>
            <pubDate>Thu, 18 Dec 2025 03:55:51 GMT</pubDate>
            <atom:updated>2025-12-18T03:55:51.016Z</atom:updated>
            <content:encoded><![CDATA[<p>SciAI in action: We are at v.0.9 of SciAI. We have uploaded 70, 000 articles and vectorized them as basis for our RAG. I will soon publish a precise description how the system works. Here is a taste of it. As we started to work about sports and PD, we might have the most comprehensive knowledge base on this topic.</p><p>The review below is fully AI produced. It is only study material. It is not peer reviewed. By no means it should be taken as medical advice!</p><h3>Literature Review: This review has one extraordinary important question: Does physical exercise slow down the development of Parkinson’s disease? it is without contradiction that sport is improving life quality and has a beneficial effect on PD symptoms. But the question in this review is hard and crude. Does it slow down the physical degeneration. If there are different findings or opinions, this should be evidenced, so research gaps. Every statement should be well documented.</h3><p><em>Based on 150 papers | Generated Wed 17 Dec 2025 05:13:55 PM CET</em></p><h3>Key Findings</h3><p><strong>Key Findings</strong></p><p>A convergent body of evidence from epidemiological studies, animal models, and clinical trials strongly suggests that physical exercise may slow the progression of Parkinson’s disease (PD), extending beyond its well-established role in improving symptoms and quality of life. Epidemiological data consistently indicate that higher levels of physical activity are associated with a reduced risk of developing PD, with a pooled hazard ratio of 0.66 when comparing the highest to lowest activity levels (PMID: 32636740). Furthermore, prospective studies in individuals already diagnosed with PD show that those reporting higher baseline physical activity experience slower rates of motor and non-motor symptom progression over time (PMID: 32636740). This is supported by biological plausibility from preclinical research, where exercise induces mechanisms conducive to neuroprotection, including reduced inflammation, decreased α-synuclein expression, and increased brain-derived neurotrophic factor (BDNF) (PMID: 32636740, PMID: 40783711).</p><p>Systematic reviews and meta-analyses of randomized controlled trials provide further, though nuanced, evidence for a disease-modifying effect. A systematic review of 40 trials concluded that exercise may slow PD progression, citing improved motor scores on the Unified Parkinson’s Disease Rating Scale (UPDRS) and increased BDNF levels as key indicators (PMID: 39456173). This aligns with findings that exercise can normalize brain activation patterns and improve dopamine receptor binding (PMID: 39456173). However, the evidence grade is often low to very low, and not all biomarkers of progression (e.g., inflammatory markers or brain volume changes) consistently respond to exercise interventions, highlighting areas requiring further investigation (PMID: 39456173).</p><p>Despite this compelling associative and mechanistic data, a definitive causal conclusion remains elusive due to significant research gaps and methodological challenges. A major limitation is the difficulty in conducting long-term trials with a true “no-exercise” control group for ethical and practical reasons, and the inherent challenge in defining and measuring “disease progression” versus symptomatic improvement (PMID: 32636740). Furthermore, the optimal exercise prescription — including type, intensity, frequency, and duration — for maximizing neuroprotective effects is not yet established (PMID: 40783711, PMID: 39456173). There is also a noted gap between evidence-based recommendations and clinical practice, with therapy and physical activity promotion being under-utilized in newly diagnosed patients (PMID: 40783711).</p><p>In summary, the literature presents a robust and consistent signal that physical activity is associated with a reduced risk of PD and potentially slower progression after diagnosis, supported by plausible neurobiological mechanisms. The consensus is that exercise provides clear symptomatic benefits and may modify disease course, but the most rigorous evidence for slowing underlying neurodegeneration is still evolving. Future research must prioritize long-term, well-controlled trials with biomarker outcomes to translate this promising association into a definitive therapeutic strategy (PMID: 32636740, PMID: 39456173).</p><h3>Contradictory Evidence</h3><p><strong>Contradictory Evidence</strong></p><p>Despite a compelling theoretical rationale and widespread clinical advocacy for exercise in Parkinson’s disease (PD), direct and consistent evidence that physical activity slows the underlying neurodegeneration — as opposed to providing symptomatic relief — remains elusive and debated. The primary contradiction lies between robust data showing symptomatic and functional benefits and the paucity of conclusive human trials demonstrating disease-modifying effects. While epidemiological studies often suggest an association between higher physical activity levels and a reduced risk of developing PD or slower symptom progression, these observational designs cannot establish causality due to potential reverse causation (i.e., reduced activity being an early prodromal symptom) and confounding variables (PMID: 32636740). Furthermore, not all cohort studies support this inverse relationship; some, such as the work by Logroscino et al. (2006) cited by Crotty and Schwarzschild, have found no significant association, highlighting inconsistency in the epidemiological foundation (PMID: 32636740).</p><p>The nature of the disagreement extends to interventional trials, where significant methodological challenges preclude definitive conclusions. Most randomized controlled trials (RCTs) are of short duration (weeks to months), focus on symptomatic endpoints, and lack a true inert control, making it impossible to disentangle neuroplastic adaptation from neuroprotection. As noted in a recent systematic review, while some trials show exercise leads to lower motor scores and improved biomarkers like brain-derived neurotrophic factor (BDNF), the evidence grade is “low to very low,” and consistent findings on neuroimaging or inflammatory markers are lacking (PMID: 39456173). This indicates that while biological plausibility exists, the translational evidence for halting physical degeneration is not yet robust. A major point of contention is the very definition of “disease modification” and the appropriate biomarkers or clinical endpoints needed to measure it, a challenge explicitly outlined in the literature (PMID: 32636740).</p><p>Key factors explaining these conflicting findings and research gaps include heterogeneous study populations, varying definitions of “exercise” (type, intensity, frequency), and the difficulty of conducting long-term, placebo-controlled trials with a behavioral intervention. The scoping review by PMID: 40783711 underscores the challenge of defining “newly diagnosed” PD and the heterogeneity in intervention approaches, which complicates the synthesis of evidence. Ultimately, while there is unanimous agreement on the symptomatic and quality-of-life benefits of exercise, the central question of whether it decelerates the core degenerative process of PD remains empirically unanswered. The current state of evidence, as summarized by Crotty and Schwarzschild, reveals a landscape of “convergent epidemiological and laboratory data” that is promising yet still requires validation through meticulously designed, long-term disease-modification trials (PMID: 32636740).</p><h3>Research Gaps</h3><p><strong>Research Gaps</strong></p><p>Despite a compelling theoretical rationale and promising observational data, definitive evidence from human trials that physical exercise slows the underlying neurodegeneration in Parkinson’s disease (PD) remains elusive. A critical gap exists in long-term, randomized controlled trials (RCTs) with disease progression as a primary outcome, as most existing trials are relatively short-term and measure symptomatic improvement rather than true disease modification (PMID: 32636740). While systematic reviews note that exercise may lower motor severity scores and improve biomarkers like brain-derived neurotrophic factor (BDNF), the overall quality of evidence is low to very low, and findings on neuroimaging markers of progression or inflammatory biomarkers are inconsistent (PMID: 39456173). This highlights a fundamental methodological gap: the lack of consensus on and validation of biomarkers that can serve as surrogate endpoints for neurodegeneration in exercise trials, beyond clinical rating scales which are susceptible to symptomatic effects.</p><p>Significant gaps also persist in defining the optimal parameters of an exercise intervention for neuroprotection. The literature underscores a need to elucidate the precise type, intensity, amount, and duration of physical activity required for a potential disease-modifying effect (PMID: 40783711, PMID: 39456173). While moderate-to-vigorous aerobic exercise is often suggested, the translationally relevant features — such as whether continuous aerobic, high-intensity interval, or complex skill-based training is superior — are not well-established. Furthermore, research has insufficiently addressed how these parameters should be tailored across disease stages, genotypes, or phenotypic variants of PD. The epidemiological finding of a potential gender difference in risk reduction, with stronger associations observed in men, further points to a gap in understanding biological and sociobehavioral modifiers of exercise’s effect (PMID: 32636740).</p><p>Finally, there is a pronounced implementation gap between evidence and practice, particularly in the crucial early stages of PD. As noted in a scoping review, therapy for newly diagnosed patients is under-utilized, with low referral rates to physiotherapy, creating a disconnect between the emphasis on early intervention and real-world clinical pathways (PMID: 40783711). This gap extends to research; most trials have been conducted in participants with established motor symptoms, leaving a paucity of evidence on the impact of initiating exercise in the <em>de novo</em> or prodromal phase. Future research must therefore not only address biological efficacy but also develop and test scalable, personalized intervention models that can be successfully integrated into healthcare systems to ensure people with PD can access and adhere to potentially neuroprotective exercise regimens from the point of diagnosis onward.</p><h3>Future Directions</h3><p><strong>Future Directions</strong></p><p>The current literature converges on the premise that physical exercise is beneficial for symptom management in Parkinson’s disease (PD) and provides compelling, though not yet definitive, evidence that it may slow disease progression. The critical question of whether exercise directly decelerates the underlying neurodegeneration — the “hard and crude” question of this review — remains partially unanswered due to methodological limitations in existing human trials. Consequently, future research must prioritize study designs capable of isolating true disease-modifying effects from symptomatic benefits. A primary direction involves conducting large-scale, long-duration randomized controlled trials (RCTs) with de novo PD patients, employing delayed-start or washout designs. These methodologies can help distinguish whether early, sustained exercise alters the trajectory of decline in a manner inconsistent with purely symptomatic effects (PMID: 32636740). Such trials must incorporate biomarkers of neuroprotection and neurodegeneration as secondary or exploratory outcomes. Promising biomarkers include neuroimaging measures of nigrostriatal integrity (e.g., dopamine transporter SPECT), levels of neurotrophic factors like brain-derived neurotrophic factor (BDNF), and inflammatory markers, which have shown sensitivity to exercise in preliminary studies (PMID: 32636740; PMID: 39456173).</p><p>A second crucial direction is the precise delineation of an “exercise prescription” for neuroprotection. Current evidence supports the general benefits of physical activity but lacks specificity regarding the optimal type, intensity, frequency, and duration required to maximize potential disease-modifying effects. As highlighted in recent reviews, research must move beyond demonstrating that exercise is beneficial to elucidating which exercise parameters are most critical (PMID: 40783711; PMID: 39456173). Comparative effectiveness trials are needed to contrast modalities such as high-intensity aerobic training, resistance exercise, and complex skill-based activities (e.g., dance, boxing) on biomarker and clinical progression outcomes. Furthermore, the interaction between exercise timing (early versus later in the disease course) and disease phenotype warrants investigation, as the neuroprotective potential may be greatest during a critical window early in the pathogenic process.</p><p>Finally, significant research gaps exist in understanding the mechanisms of exercise-induced neuroprotection and translating these into accessible interventions. Future work should integrate basic science on exercise-mediated pathways — such as upregulation of neurotrophic factors, enhanced mitochondrial function, reduced inflammation, and modulation of α-synuclein pathology — with clinical studies (PMID: 32636740; PMID: 39456173). Concurrently, implementation science is needed to address the documented gap between evidence and practice, where only a minority of newly diagnosed patients are referred to physiotherapy (PMID: 40783711). Research should focus on developing and testing scalable, personalized intervention models that promote long-term adherence, potentially leveraging telehealth and community-based programs. By combining mechanistic studies, precision exercise trials, and implementation research, the field can advance from affirming the value of exercise to defining how it can be reliably harnessed to modify the progression of Parkinson’s disease.</p><h3>References</h3><ol><li>Gilby, Jonathan, Kent, Bridie, Lozano, Rachel Knight, Marsden, Jonathan (2025) <em>Interventions for promoting physical activity in people with newly diagnosed Parkinson’s disease: scoping review.</em> <strong>Syst Rev</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40783711/">PubMed</a> | <a href="https://doi.org/10.1186/s13643-025-02892-2">Full Text</a></li><li>Grace F Crotty, Michael A Schwarzschild (2020) <em>Chasing Protection in Parkinson’s Disease: Does Exercise Reduce Risk and Progression?</em> <strong>Frontiers in aging neuroscience</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32636740/">PubMed</a> | <a href="https://doi.org/10.3389/fnagi.2020.00186">Full Text</a></li><li>Mitchell, Alexandra K, Bliss, Rebecca R, Church, Frank C (2024) <em>Exercise, Neuroprotective Exerkines, and Parkinson’s Disease: A Narrative Review.</em> <strong>Biomolecules</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39456173/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11506540">Full Text</a></li><li>Gronek et al. (2021) <em>The Mechanism of Physical Activity-induced Amelioration of Parkinson’s Disease: A Narrative Review.</em> <strong>Aging Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33532136/">PubMed</a> | <a href="https://doi.org/10.14336/AD.2020.0407">Full Text</a></li><li>Padilha et al. (2023) <em>Physical exercise and its effects on people with Parkinson’s disease: Umbrella review.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37917715/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0293826">Full Text</a></li><li>Paterno, Andrea, Polsinelli, Giovanni, Federico, Bruno (2024) <em>Changes of brain-derived neurotrophic factor (BDNF) levels after different exercise protocols: a systematic review of clinical studies in Parkinson’s disease.</em> <strong>Front Physiol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38444767/">PubMed</a> | <a href="https://doi.org/10.3389/fphys.2024.1352305">Full Text</a></li><li>Ketaki S Bhalsing, Louis C S Tan, Masoom M Abbas (2018) <em>Role of Physical Activity in Parkinson’s Disease.</em> <strong>Annals of Indian Academy of Neurology</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/30532351/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6238554">Full Text</a></li><li>Mahalakshmi, B, Maurya, Nancy, Lee, Shin-Da, Bharath Kumar, V (2020) <em>Possible Neuroprotective Mechanisms of Physical Exercise in Neurodegeneration.</em> <strong>Int J Mol Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32824367/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7460620">Full Text</a></li><li>Langeskov-Christensen, Martin, Franzén, Erika, Grøndahl Hvid, Lars, Dalgas, Ulrik (2024) <em>Exercise as medicine in Parkinson’s disease.</em> <strong>J Neurol Neurosurg Psychiatry</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38418216/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/38418216">Full Text</a></li><li>Ernst et al. (2024) <em>Physical exercise for people with Parkinson’s disease: a systematic review and network meta-analysis.</em> <strong>Cochrane Database Syst Rev</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38588457/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/38588457">Full Text</a></li><li>Li et al. (2023) <em>Does Exercise Attenuate Disease Progression in People With Parkinson’s Disease? A Systematic Review With Meta-Analyses.</em> <strong>Neurorehabil Neural Repair</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37166181/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/10272626">Full Text</a></li><li>Lynn Rochester, Terry Ellis (2018) <em>Mobilizing Parkinson’s Disease: The Future of Exercise.</em> <strong>Journal of Parkinson’s disease</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/30584167/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6311359">Full Text</a></li><li>Ana Castro Caldas et al. (2020) <em>Physical Activity, Exercise, and Physiotherapy in Parkinson’s Disease: Defining the Concepts.</em> <strong>Movement disorders clinical practice</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/31970204/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6962680">Full Text</a></li><li>Fan et al. (2020) <em>What and How Can Physical Activity Prevention Function on Parkinson’s Disease?</em> <strong>Oxid Med Cell Longev</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32215173/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7042542/">Full Text</a></li><li>Emig, Mallory, George, Tikku, Zhang, Justin K, Soudagar-Turkey, Momina (2021) <em>The Role of Exercise in Parkinson’s Disease.</em> <strong>J Geriatr Psychiatry Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34219524/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/34219524">Full Text</a></li><li>Chromiec, Patryk Andrzej, Urbaś, Zofia Kinga, Jacko, Martyna, Kaczor, Jan Jacek (2021) <em>The Proper Diet and Regular Physical Activity Slow Down the Development of Parkinson Disease.</em> <strong>Aging Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34631210/">PubMed</a> | <a href="https://doi.org/10.14336/AD.2021.0123">Full Text</a></li><li>Nay et al. (2021) <em>Molecular Mechanisms Underlying the Beneficial Effects of Exercise on Brain Function and Neurological Disorders.</em> <strong>Int J Mol Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33919972/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8070923">Full Text</a></li><li>McGinley, Jennifer L, Nakayama, Yasuhide (2024) <em>Exercise for People with Parkinson’s Disease: Updates and Future Considerations.</em> <strong>Phys Ther Res</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39257520/">PubMed</a> | <a href="https://doi.org/10.1298/ptr.R0030">Full Text</a></li><li>Diaz-Galvan et al. (2025) <em>Association of Physical Exercise With Structural Brain Changes and Cognitive Decline in Patients With Early Parkinson Disease.</em> <strong>Neurology</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40768688/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/40768688">Full Text</a></li><li>Hao, Zikang, Zhang, Xiaodan, Chen, Ping (2022) <em>Effects of Ten Different Exercise Interventions on Motor Function in Parkinson’s Disease Patients-A Network Meta-Analysis of Randomized Controlled Trials.</em> <strong>Brain Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35741584/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9221238">Full Text</a></li><li>Megan Lee, Pei-Ling Wu, Tzu-Ting Huang (2017) <em>Effectiveness of physical activity on patients with depression and Parkinson’s disease: A systematic review.</em> <strong>PloS one</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/28749970/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0181515">Full Text</a></li><li>Benedetta Bernazzoli, Cosimo Costantino, Ileana Ramazzina (2017) <em>Systematic review on strength training in Parkinson’s disease: an unsolved question.</em> <strong>Clinical interventions in aging</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/28408811/">PubMed</a> | <a href="https://doi.org/10.2147/CIA.S131903">Full Text</a></li><li>Caroline M Tanner, Cynthia L Comella, Sara C LaHue (2016) <em>The best medicine? The influence of physical activity and inactivity on Parkinson’s disease.</em> <strong>Movement disorders : official journal of the Movement Disorder Society</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/27477046/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9491025">Full Text</a></li><li>Silva, S V, Moreira, G M S, Campos-Junior, P H A, Damázio, L C M (2024) <em>The benefits of high-intensity physical exercise before and after Parkinson’s disease induction in rats.</em> <strong>Braz J Biol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39417437/">PubMed</a> | <a href="https://doi.org/10.1590/1519-6984.282438">Full Text</a></li><li>Li, Ying, Zhuang, Ruixin, Zhang, Jianhua, Liu, Xiaojie (2025) <em>The effect of different exercise training modes on improving quality of life in patients with Parkinson’s disease: a network analysis.</em> <strong>Front Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40672448/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12264356/">Full Text</a></li><li>Chunhui Zhao, Dan Han, Fudi Wang, Junxia Min, Pan Zhang, Qi Cheng, Xuexian Fang (2018) <em>Association of Levels of Physical Activity With Risk of Parkinson Disease: A Systematic Review and Meta-analysis.</em> <strong>JAMA network open</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/30646166/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6324511">Full Text</a></li><li>Paillard, Thierry, Rolland, Yves, de Souto Barreto, Philipe (2015) <em>Protective Effects of Physical Exercise in Alzheimer’s Disease and Parkinson’s Disease: A Narrative Review.</em> <strong>J Clin Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/26174783/">PubMed</a> | <a href="https://doi.org/10.3988/jcn.2015.11.3.212">Full Text</a></li><li>Ellis et al. (2021) <em>Evidence for Early and Regular Physical Therapy and Exercise in Parkinson’s Disease.</em> <strong>Semin Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33742432/">PubMed</a> | <a href="https://doi.org/10.1055/s-0041-1725133">Full Text</a></li><li>Oosterhof, Thomas H, Darweesh, Sirwan K L, Bloem, Bastiaan R, de Vries, Nienke M (2024) <em>Considerations on How to Prevent Parkinson’s Disease Through Exercise.</em> <strong>J Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39031383/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11492051">Full Text</a></li><li>Li, Wei-Qi, Yang, Jia-Hua, Liu, Lu-Lu, Ding, Song-Tao, Yu, Bin, Jiang, Lin, Yan, Ning, Liu, Han-Deng (2025) <em>Exercise as a multitarget therapy: modulating myokines, neurotrophins, and inflammation in Parkinson’s disease.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40933823/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12417431/">Full Text</a></li><li>Yang et al. (2022) <em>Efficacy and evaluation of therapeutic exercises on adults with Parkinson’s disease: a systematic review and network meta-analysis.</em> <strong>BMC Geriatr</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36271367/">PubMed</a> | <a href="https://doi.org/10.1186/s12877-022-03510-9">Full Text</a></li><li>Christian Duval, Jean-Francois Daneault, Martine Lauzé (2016) <em>The Effects of Physical Activity in Parkinson’s Disease: A Review.</em> <strong>Journal of Parkinson’s disease</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/27567884/">PubMed</a> | <a href="http://nrs.harvard.edu/urn-3:HUL.InstRepos:29626010">Full Text</a></li><li>Liu, Zhaoyang, Lemus, Jessica, Smirnova, Irina V, Liu, Wen (2023) <em>Rehabilitation for non-motor symptoms for patients with Parkinson’s disease from an α-synuclein perspective: a narrative review.</em> <strong>Explor Neuroprotective Ther</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37920444/">PubMed</a> | <a href="https://www.explorationpub.com/uploads/Article/A100449/100449.pdf">Full Text</a></li><li>Kim et al. (2025) <em>Motivators and Barriers Affecting Exercise in Patients With Parkinson’s Disease.</em> <strong>J Clin Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39778563/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11711269/">Full Text</a></li><li>Tsai, Chia-Liang, Chien, Chung-Yao, Pan, Chien-Yu, Tseng, Yu-Ting, Wang, Tsai-Chiao, Lin, Tsu-Kung (2025) <em>Effects of long-term Tai Chi vs. aerobic exercise on antioxidant activity and cognitive function in individuals with Parkinson’s disease.</em> <strong>Behav Brain Res</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39332640/">PubMed</a> | <a href="https://doi.org/10.1016/j.bbr.2024.115274">Full Text</a></li><li>J McClelland, M E Morris, P Alves Da Rocha (2015) <em>Complementary physical therapies for movement disorders in Parkinson’s disease: a systematic review.</em> <strong>European journal of physical and rehabilitation medicine</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/26138090/">PubMed</a></li><li>Müller, Jan, Myers, Jonathan (2018) <em>Association between physical fitness, cardiovascular risk factors, and Parkinson’s disease.</em> <strong>Eur J Prev Cardiol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/29667433/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/29667433">Full Text</a></li><li>Dommershuijsen et al. (2023) <em>The elephant in the room: critical reflections on mortality rates among individuals with Parkinson’s disease.</em> <strong>NPJ Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37857675/">PubMed</a> | <a href="https://www.nature.com/articles/s41531-023-00588-9.pdf">Full Text</a></li><li>Duarte et al. (2023) <em>Physical activity based on dance movements as complementary therapy for Parkinson’s disease: Effects on movement, executive functions, depressive symptoms, and quality of life.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36730266/">PubMed</a> | <a href="https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0281204&amp;type=printable">Full Text</a></li><li>Magaña, Juan Carlos, Deus, Cláudia M, Giné-Garriga, Maria, Montané, Joel, Pereira, Susana P (2022) <em>Exercise-Boosted Mitochondrial Remodeling in Parkinson’s Disease.</em> <strong>Biomedicines</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36551984/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9775656">Full Text</a></li><li>Tuijt et al. (2020) <em>Self-Management Components as Experienced by People with Parkinson’s Disease and Their Carers: A Systematic Review and Synthesis of the Qualitative Literature.</em> <strong>Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33489082/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7787805">Full Text</a></li><li>Alnes et al. (2023) <em>Mobile health technology, exercise adherence and optimal nutrition post rehabilitation among people with Parkinson’s Disease (mHEXANUT) — a randomized controlled trial protocol.</em> <strong>BMC Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36864377/">PubMed</a> | <a href="https://bmcneurol.biomedcentral.com/counter/pdf/10.1186/s12883-023-03134-5">Full Text</a></li><li>Su, Yiping, Su, Zhanguo (2025) <em>Effects of exercise on neuroinflammation in age-related neurodegenerative disorders.</em> <strong>Eur J Med Res</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/41024269/">PubMed</a> | <a href="https://eurjmedres.biomedcentral.com/counter/pdf/10.1186/s40001-025-03165-3">Full Text</a></li><li>Zhen, Kai, Zhang, Shiyan, Tao, Xifeng, Li, Gen, Lv, Yuanyuan, Yu, Laikang (2022) <em>A systematic review and meta-analysis on effects of aerobic exercise in people with Parkinson’s disease.</em> <strong>NPJ Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36316416/">PubMed</a> | <a href="https://doi.org/10.1038/s41531-022-00418-4">Full Text</a></li><li>Camila Torriani-Pasin et al. (2024) <em>Factors Associated With Physical Activity and Sedentary Behavior in People With Parkinson Disease: A Systematic Review and Meta-Analysis.</em> <strong>Physical therapy</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39126377/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/39126377">Full Text</a></li><li>Ernst et al. (2023) <em>Physical exercise for people with Parkinson’s disease: a systematic review and network meta-analysis.</em> <strong>Cochrane Database Syst Rev</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36602886/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9815433">Full Text</a></li><li>Ahmad, Syed O, Stiles, Dana, Brown, Kaylee, Dillon, Leah, Shroba, Eliza (2025) <em>Meta-Analysis of Exercise Effects on Cognition in Persons with Parkinson’s Disease.</em> <strong>NeuroSci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40559207/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12195965/">Full Text</a></li><li>Jagota et al. (2024) <em>Exercise Habits in People with Parkinson’s: A Multinational Survey.</em> <strong>Mov Disord Clin Pract</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39206964/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11542280">Full Text</a></li><li>Alizadeh et al. (2025) <em>EEG dynamical features during variable-intensity cycling exercise in Parkinson’s disease.</em> <strong>Front Hum Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40356882/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12066486/">Full Text</a></li><li>Ji, Xiaotian, Lu, Danian, Yang, Qinglan, Xiao, Linting, Wang, Jing, Wang, Gaiqing (2022) <em>Physical Therapy for at Least 6 Months Improves Motor Symptoms in Parkinson’s Patients: A Meta-Analysis.</em> <strong>Comput Math Methods Med</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35959355/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9357765">Full Text</a></li><li>Lee, Dae-Hwan, Woo, Bong-Sik, Park, Yong-Hwa, Lee, Jung-Ho (2024) <em>General Treatments Promoting Independent Living in Parkinson’s Patients and Physical Therapy Approaches for Improving Gait-A Comprehensive Review.</em> <strong>Medicina (Kaunas)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38792894/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11123276">Full Text</a></li><li>Kamo, Hikaru, Oyama, Genko, Nagaki, Koichi, Hattori, Nobutaka (2025) <em>Quantification of rehabilitation for patients with Parkinson’s disease and related disorders using wearable devices; a proof of concept.</em> <strong>Front Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40771968/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12325046/">Full Text</a></li><li>Khan, Aiza, Ezeugwa, Joy, Ezeugwu, Victor E (2024) <em>A systematic review of the associations between sedentary behavior, physical inactivity, and non-motor symptoms of Parkinson’s disease.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38551932/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0293382">Full Text</a></li><li>Machado et al. (2022) <em>Clinical applications of exercise in Parkinson’s disease: what we need to know?</em> <strong>Expert Rev Neurother</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36168890/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/36168890">Full Text</a></li><li>Luthra et al. (2025) <em>Aerobic exercise-induced changes in fluid biomarkers in Parkinson’s disease.</em> <strong>NPJ Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40595707/">PubMed</a> | <a href="https://www.nature.com/articles/s41531-025-01042-8.pdf">Full Text</a></li><li>Schootemeijer, Sabine, van der Kolk, Nicolien M, Bloem, Bastiaan R, de Vries, Nienke M (2020) <em>Current Perspectives on Aerobic Exercise in People with Parkinson’s Disease.</em> <strong>Neurotherapeutics</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32808252/">PubMed</a> | <a href="https://link.springer.com/content/pdf/10.1007/s13311-020-00904-8.pdf">Full Text</a></li><li>Morley et al. (2025) <em>Physical Activity, Patient-Reported Outcomes, and Quality of Life in Parkinson’s Disease.</em> <strong>J Geriatr Psychiatry Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40444999/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/12640359">Full Text</a></li><li>Ascherio, Alberto, Schwarzschild, Michael A (2016) <em>The epidemiology of Parkinson’s disease: risk factors and prevention.</em> <strong>Lancet Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/27751556/">PubMed</a> | <a href="https://doi.org/10.1016/s1474-4422(16)30230-7">Full Text</a></li><li>Tiihonen, Marianne, Westner, Britta U, Butz, Markus, Dalal, Sarang S (2021) <em>Parkinson’s disease patients benefit from bicycling — a systematic review and meta-analysis.</em> <strong>NPJ Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34561455/">PubMed</a> | <a href="https://doi.org/10.1038/s41531-021-00222-6">Full Text</a></li><li>Jun Feng Teng, Shu Ji Jiao, Wei Qing Qiu, Yan Lei Hao, Yan Yang, Zhan Yun Lv (2015) <em>The efficacy of traditional Chinese Medical Exercise for Parkinson’s disease: a systematic review and meta-analysis.</em> <strong>PloS one</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/25830664/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0122469">Full Text</a></li><li>Małczyńska-Sims, Paulina, Chalimoniuk, Małgorzata, Sułek, Anna (2020) <em>The Effect of Endurance Training on Brain-Derived Neurotrophic Factor and Inflammatory Markers in Healthy People and Parkinson’s Disease. A Narrative Review.</em> <strong>Front Physiol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33329027/">PubMed</a> | <a href="https://doi.org/10.3389/fphys.2020.578981">Full Text</a></li><li>Thacker et al. (2008) <em>Recreational physical activity and risk of Parkinson’s disease.</em> <strong>Mov Disord</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/17960818/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2387117/">Full Text</a></li><li>Zheng, Yalin, Shen, Yu, Feng, Renzhi, Hu, Weiyin, Huang, Peng (2024) <em>Research progress on the application of anti-gravity treadmill in the rehabilitation of Parkinson’s disease patients: a mini review.</em> <strong>Front Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38882698/">PubMed</a> | <a href="https://doi.org/10.3389/fneur.2024.1401256">Full Text</a></li><li>van der Berg, Indy, Schootemeijer, Sabine, Overbeek, Karin, Bloem, Bastiaan R, de Vries, Nienke M (2024) <em>Dietary Interventions in Parkinson’s Disease.</em> <strong>J Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38277304/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/10836553">Full Text</a></li><li>Moratelli et al. (2025) <em>Functional Training and Mat Pilates Are Effective in Improving Cardiorespiratory Capacity and Strength in People With Parkinson’s Disease: A Randomized Clinical Trial With Follow-up.</em> <strong>J Strength Cond Res</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40570285/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/40570285">Full Text</a></li><li>Zhang, Qilu, Shi, Mengyao, Zhang, Jing, Yin, Xiangyan, Chen, Yi, Wang, Xiaoxiao, Zhang, Yonghong (2025) <em>Association between Physical Activity and Parkinson’s Disease: A Prospective Cohort Study.</em> <strong>Neuroepidemiology</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39038446/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/39038446">Full Text</a></li><li>Seemiller, Joseph, Mahajan, Abhimanyu, Morrow, Christopher B, Pontone, Gregory M, Mills, Kelly A (2025) <em>Does exercise demonstrate a dose-dependent benefit with cognition in Parkinson’s disease?</em> <strong>Parkinsonism Relat Disord</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39848137/">PubMed</a> | <a href="https://doi.org/10.1016/j.parkreldis.2025.107277">Full Text</a></li><li>Yoon, Seo Yeon (2022) <em>Update on Parkinson’s Disease Rehabilitation.</em> <strong>Brain Neurorehabil</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36743207/">PubMed</a> | <a href="https://doi.org/10.12786/bn.2022.15.e15">Full Text</a></li><li>Schootemeijer et al. (2025) <em>Promoting Physical Activity in People With Parkinson’s Disease Through a Smartphone App: A Pilot Study.</em> <strong>J Neurol Phys Ther</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39763021/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/39763021">Full Text</a></li><li>Yang et al. (2022) <em>Effects and parameters of community-based exercise on motor symptoms in Parkinson’s disease: a meta-analysis.</em> <strong>BMC Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36581847/">PubMed</a> | <a href="https://doi.org/10.1186/s12883-022-03027-z">Full Text</a></li><li>Baron, Fred, Bogosian, Angeliki (2025) <em>Exploring social, cultural and environmental factors that influence attitudes to exercise among people with Parkinson’s disease: A qualitative study.</em> <strong>J Health Psychol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39575978/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/12381388">Full Text</a></li><li>Tumas et al. (2025) <em>Modifiable risk factors associated with the risk of developing Parkinson’s disease: a critical review.</em> <strong>Arq Neuropsiquiatr</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40107260/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11922612/">Full Text</a></li><li>A Bernetti et al. (2020) <em>Evidence of Rehabilitative Impact of Progressive Resistance Training (PRT) Programs in Parkinson Disease: An Umbrella Review.</em> <strong>Parkinson’s disease</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32566123/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7270996">Full Text</a></li><li>Marta F D Trindade, Rui A Viana (2021) <em>Effects of auditory or visual stimuli on gait in Parkinsonic patients: a systematic review.</em> <strong>Porto biomedical journal</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34368491/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8341369">Full Text</a></li><li>Costa et al. (2024) <em>Physical Exercise for Treating the Anxiety and Depression Symptoms of Parkinson’s Disease: Systematic Review and Meta-Analysis.</em> <strong>J Geriatr Psychiatry Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38445606/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/38445606">Full Text</a></li><li>Still, Amanda, Hale, Leigh, Alam, Sarfaraz, Morris, Meg E, Jayakaran, Prasath (2024) <em>Relationships between physical activities performed under free-living conditions and non-motor symptoms in people with Parkinson’s: A systematic review and meta-analysis.</em> <strong>Clin Rehabil</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39175369/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11528973">Full Text</a></li><li>Wilson, Alice C, Pountney, Dean L, Khoo, Tien K (2025) <em>Therapeutic Mechanisms of Exercise in Parkinson’s Disease.</em> <strong>Int J Mol Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40429998/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12112306/">Full Text</a></li><li>Pinto et al. (2024) <em>Quality assessment and umbrella review of systematic reviews about dance for people with Parkinson’s disease.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39739959/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0311003">Full Text</a></li><li>Rafferty et al. (2024) <em>Stemming the Tide: The Proactive Role of Allied Health Therapy in Parkinson’s Disease.</em> <strong>J Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38848194/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11380284">Full Text</a></li><li>Colum D MacKinnon, Michael J Howell, Miriam R Rafferty, Rebekah L S Summers (2021) <em>Motor Dysfunction in REM Sleep Behavior Disorder: A Rehabilitation Framework for Prodromal Synucleinopathy.</em> <strong>Neurorehabilitation and neural repair</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33978530/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8225559">Full Text</a></li><li>Erwin Eh van Wegen, Gert Kwakkel, Henk W Berendse, John Verhoef, Roy G Elbers (2015) <em>Interventions for fatigue in Parkinson’s disease.</em> <strong>The Cochrane database of systematic reviews</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/26447539/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9240814">Full Text</a></li><li>Muleiro Alvarez et al. (2024) <em>A Comprehensive Approach to Parkinson’s Disease: Addressing Its Molecular, Clinical, and Therapeutic Aspects.</em> <strong>Int J Mol Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39000288/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11241043">Full Text</a></li><li>Li, Zhiqin, Hu, Pingqing (2025) <em>Exercise: The key to enhancing sleep quality and physical function in Parkinson’s disease? A systematic review and meta-analysis.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/41296768/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0336381">Full Text</a></li><li>Farashi, Sajjad, Kiani, Leila, Bashirian, Saeid (2021) <em>Effect of Exercise on Parkinson’s Disease Tremor: A Meta-analysis Study.</em> <strong>Tremor Other Hyperkinet Mov (N Y)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33981478/">PubMed</a> | <a href="https://doi.org/10.5334/tohm.599">Full Text</a></li><li>Franzén et al. (2019) <em>The EXPANd trial: effects of exercise and exploring neuroplastic changes in people with Parkinson’s disease: a study protocol for a double-blinded randomized controlled trial.</em> <strong>BMC Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/31718583/">PubMed</a> | <a href="https://bmcneurol.biomedcentral.com/track/pdf/10.1186/s12883-019-1520-2">Full Text</a></li><li>Santiago, Jose A, Potashkin, Judith A (2023) <em>Physical activity and lifestyle modifications in the treatment of neurodegenerative diseases.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37304072/">PubMed</a> | <a href="https://www.frontiersin.org/articles/10.3389/fnagi.2023.1185671/pdf">Full Text</a></li><li>Beomseok Jeon, Do Kyung Ko, Hanall Lee, Nyeonju Kang, Ryul Kim, Tae Lee Lee (2023) <em>Effects of Exercise on Depressive Symptoms in Patients With Parkinson Disease: A Meta-analysis.</em> <strong>Neurology</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36220597/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/36220597">Full Text</a></li><li>Jeann L Sabino-Carvalho, Lauro C Vianna (2020) <em>Altered cardiorespiratory regulation during exercise in patients with Parkinson’s disease: A challenging non-motor feature.</em> <strong>SAGE open medicine</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32435491/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7222646">Full Text</a></li><li>Llamas-Velasco et al. (2021) <em>Physical activity and risk of Parkinson’s disease and parkinsonism in a prospective population-based study (NEDICES).</em> <strong>Prev Med Rep</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34307002/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8283037/">Full Text</a></li><li>KSNR PD Writing Group et al. (2020) <em>KSNR Clinical Consensus Statements: Rehabilitation of Patients with Parkinson’s Disease.</em> <strong>Brain Neurorehabil</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36744191/">PubMed</a> | <a href="http://e-bnr.org/Synapse/Data/PDFData/0176BN/bn-13-e17.pdf">Full Text</a></li><li>Zhou, Xiao, Zhao, Peng, Guo, Xuanhui, Wang, Jialin, Wang, Ruirui (2022) <em>Effectiveness of aerobic and resistance training on the motor symptoms in Parkinson’s disease: Systematic review and network meta-analysis.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35978948/">PubMed</a> | <a href="https://doi.org/10.3389/fnagi.2022.935176">Full Text</a></li><li>Kim, Younguk, Kim, Jin Hyun, Ridgel, Angela L (2025) <em>Impact of Physical Activity Levels on Parkinson’s Disease Motor and Nonmotor Symptoms and Quality of Life in Older Adults With Parkinson’s Disease.</em> <strong>J Aging Phys Act</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40335159/">PubMed</a> | <a href="https://doi.org/10.1123/japa.2024-0252">Full Text</a></li><li>Reichmann et al. (2022) <em>Life style and Parkinson’s disease.</em> <strong>J Neural Transm (Vienna)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35606622/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9463300/">Full Text</a></li><li>Colleen G Canning, Natalie E Allen, Niamh Moloney, Vanessa van Vliet (2015) <em>The Rationale for Exercise in the Management of Pain in Parkinson’s Disease.</em> <strong>Journal of Parkinson’s disease</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/25649828/">PubMed</a> | <a href="https://content.iospress.com/download/journal-of-parkinsons-disease/jpd140508?id=journal-of-parkinsons-disease%2Fjpd140508">Full Text</a></li><li>Suárez-Iglesias, David, Miller, Kyle J, Seijo-Martínez, Manuel, Ayán, Carlos (2019) <em>Benefits of Pilates in Parkinson’s Disease: A Systematic Review and Meta-Analysis.</em> <strong>Medicina (Kaunas)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/31412676/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6723274">Full Text</a></li><li>Lee, Myung Jun, Park, Jinse, Ryu, Dong-Woo, Yoo, Dallah, Cheon, Sang-Myung (2024) <em>Survey of the Knowledge, Attitudes, and Practices of Neurologists Regarding Exercise in Parkinson’s Disease.</em> <strong>J Clin Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38627229/">PubMed</a> | <a href="https://doi.org/10.3988/jcn.2023.0344">Full Text</a></li><li>Oosterhof, Thomas H, Schootemeijer, Sabine, de Vries, Nienke M (2023) <em>Clinical Trial Highlights — Interventions Promoting Physical Activity in Parkinson’s Disease.</em> <strong>J Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37125564/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/10200135">Full Text</a></li><li>Hartmann-Nardin, Daniel, Stock, Stephanie, Kalbe, Elke, Folkerts, Ann-Kristin (2024) <em>Cost-Effectiveness Analyses of Non-Pharmacological and Non-Surgical Interventions in Idiopathic Parkinson’s Disease: A Systematic Review.</em> <strong>J Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38339939/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11380296">Full Text</a></li><li>Uhrbrand, Anders, Stenager, Egon, Pedersen, Martin Sloth, Dalgas, Ulrik (2015) <em>Parkinson’s disease and intensive exercise therapy — a systematic review and meta-analysis of randomized controlled trials.</em> <strong>J Neurol Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/25936252/">PubMed</a> | <a href="https://doi.org/10.1016/j.jns.2015.04.004">Full Text</a></li><li>Alexander Storch, Bernhard Elsner, Jan Mehrholz, Joachim Kugler, Kathleen Hirsch, Marcus Pohl (2015) <em>Treadmill training for patients with Parkinson’s disease.</em> <strong>The Cochrane database of systematic reviews</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/26363646/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/9579750">Full Text</a></li><li>Chortane, Oussama Gaied, Magtouf, Elmoetez, Maktouf, Wael, Chortane, Sabri Gaied (2024) <em>Effects of long-term adapted physical training on functional capacity and quality of life in older adults with Parkinson’s disease.</em> <strong>AIMS Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39801793/">PubMed</a> | <a href="http://doi.org/10.3934/neuroscience.2024028">Full Text</a></li><li>Hugo Olmedillas et al. (2021) <em>Vigorous Aerobic Exercise in the Management of Parkinson Disease: A Systematic Review.</em> <strong>PM &amp; R : the journal of injury, function, and rehabilitation</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32978867/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/32978867">Full Text</a></li><li>Jopowicz, Anna, Wiśniowska, Justyna, Tarnacka, Beata (2022) <em>Cognitive and Physical Intervention in Metals’ Dysfunction and Neurodegeneration.</em> <strong>Brain Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35326301/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8946530">Full Text</a></li><li>Da Silva et al. (2021) <em>Expression levels of specific microRNAs are increased after exercise and are associated with cognitive improvement in Parkinson’s disease.</em> <strong>Mol Med Rep</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34184078/">PubMed</a> | <a href="https://www.spandidos-publications.com/10.3892/mmr.2021.12257/download">Full Text</a></li><li>Sena et al. (2023) <em>Feasibility and effect of high-intensity training on the progression of motor symptoms in adult individuals with Parkinson’s disease: A systematic review and meta-analysis.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37948405/">PubMed</a> | <a href="https://doi.org/10.1371/journal.pone.0293357">Full Text</a></li><li>Cascaes da Silva et al. (2016) <em>Effects of Physical-Exercise-Based Rehabilitation Programs on the Quality of Life of Patients With Parkinson’s Disease: A Systematic Review of Randomized Controlled Trials.</em> <strong>J Aging Phys Act</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/26751626/">PubMed</a> | <a href="https://doi.org/10.1123/japa.2015-0162">Full Text</a></li><li>Chad Swank, Samantha Cleveland, Simon Driver, Staci Shearin (2017) <em>Auditing the Physical Activity and Parkinson Disease Literature Using the Behavioral Epidemiologic Framework.</em> <strong>PM &amp; R : the journal of injury, function, and rehabilitation</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/27777097/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/27777097">Full Text</a></li><li>Johansson, Hanna, Hagströmer, Maria, Grooten, Wilhelmus J A, Franzén, Erika (2020) <em>Exercise-Induced Neuroplasticity in Parkinson’s Disease: A Metasynthesis of the Literature.</em> <strong>Neural Plast</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32256559/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7079218">Full Text</a></li><li>Byron Lai et al. (2019) <em>Exercise Training Guidelines for Multiple Sclerosis, Stroke, and Parkinson Disease: Rapid Review and Synthesis.</em> <strong>American journal of physical medicine &amp; rehabilitation</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/30844920/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6586489">Full Text</a></li><li>Alice Nieuwboer et al. (2020) <em>Barriers and Motivators to Engage in Exercise for Persons with Parkinson’s Disease.</em> <strong>Journal of Parkinson’s disease</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32925106/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7739964">Full Text</a></li><li>Jo et al. (2025) <em>Physiological change of striatum and ventral midbrain’s glia cell in response to different exercise modalities.</em> <strong>Behav Brain Res</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39571940/">PubMed</a> | <a href="https://doi.org/10.1016/j.bbr.2024.115342">Full Text</a></li><li>Svensson, Martina, Lexell, Jan, Deierborg, Tomas (2015) <em>Effects of Physical Exercise on Neuroinflammation, Neuroplasticity, Neurodegeneration, and Behavior: What We Can Learn From Animal Models in Clinical Settings.</em> <strong>Neurorehabil Neural Repair</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/25527485/">PubMed</a> | <a href="https://journals.sagepub.com/doi/pdf/10.1177/1545968314562108">Full Text</a></li><li>Lei et al. (2022) <em>The effects of different types of Tai Chi exercises on motor function in patients with Parkinson’s disease: A network meta-analysis.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36105909/">PubMed</a> | <a href="https://doi.org/10.3389/fnagi.2022.936027">Full Text</a></li><li>Zikereya, Talifu, Shi, Kaixuan, Chen, Wei (2023) <em>Goal-directed and habitual control: from circuits and functions to exercise-induced neuroplasticity targets for the treatment of Parkinson’s disease.</em> <strong>Front Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37881310/">PubMed</a> | <a href="https://doi.org/10.3389/fneur.2023.1254447">Full Text</a></li><li>Sujkowski, Alyson, Hong, Luke, Wessells, R J, Todi, Sokol V (2022) <em>The protective role of exercise against age-related neurodegeneration.</em> <strong>Ageing Res Rev</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34923167/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8761166">Full Text</a></li><li>Zhu et al. (2023) <em>Efficacy of Tai Chi on lower limb function of Parkinson’s disease patients: A systematic review and meta-analysis.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36819715/">PubMed</a> | <a href="https://doi.org/10.3389/fnagi.2023.1096417">Full Text</a></li><li>Wang, Di, Cui, Wen J, Hou, Zhen H, Gao, Ying (2023) <em>Effectiveness of different exercises in improving postural balance among Parkinson’s disease patients: a systematic review and network meta-analysis.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/37529009/">PubMed</a> | <a href="https://doi.org/10.3389/fnagi.2023.1215495">Full Text</a></li><li>De Pandis et al. (2025) <em>Mechanical peripheral stimulation for the treatment of gait disorders in patients with Parkinson’s disease: a multi-centre, double-blind, crossover randomized controlled trial.</em> <strong>J Neuroeng Rehabil</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40200338/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11978199/">Full Text</a></li><li>Shulman et al. (2013) <em>Randomized clinical trial of 3 types of physical exercise for patients with Parkinson disease.</em> <strong>JAMA Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/23128427/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/4574905">Full Text</a></li><li>Bouhadoun, Sarah, Delva, Aline, Schwarzschild, Michael A, Postuma, Ronald B (2025) <em>Preparing for Parkinson’s disease prevention trials: Current progress and future directions.</em> <strong>J Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40289581/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/40289581">Full Text</a></li><li>Rungruangbaiyok et al. (2025) <em>Effectiveness of physical therapy for locomotive syndrome: A systematic review and meta-analysis protocol.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40811515/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/12352771">Full Text</a></li><li>Wang et al. (2025) <em>Effects of boxing exercise in people with Parkinson’s disease: a systematic review.</em> <strong>Front Aging Neurosci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40212568/">PubMed</a> | <a href="https://doi.org/10.3389/fnagi.2025.1505326">Full Text</a></li><li>Ren, Hao, Zhou, Yilun, Lv, Yuanyuan, Liu, Xiaojie, He, Lingxiao, Yu, Laikang (2025) <em>Effects of Aerobic Exercise on Depressive Symptoms in People with Parkinson’s Disease: A Systematic Review and Meta-Analysis of Randomized Controlled Trials.</em> <strong>Brain Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40867124/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/40867124">Full Text</a></li><li>Fogaça et al. (2021) <em>Mind-Body Therapies From Traditional Chinese Medicine: Evidence Map.</em> <strong>Front Public Health</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34988045/">PubMed</a> | <a href="https://doi.org/10.3389/fpubh.2021.659075">Full Text</a></li><li>Chen, Yumei, Chen, Yanbin (2025) <em>The impact of combined aerobic and resistance exercise on the prognosis of early Parkinson’s disease patients.</em> <strong>Technol Health Care</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39177624/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/39177624">Full Text</a></li><li>Fekonja, Zvonka, Irgolič, Nadja, Vrbnjak, Dominika (2024) <em>Family members’ experiences of everyday caregiving for a family member living with Parkinson’s disease: a qualitative thematic analysis study.</em> <strong>BMC Nurs</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38321424/">PubMed</a> | <a href="https://bmcnurs.biomedcentral.com/counter/pdf/10.1186/s12912-024-01767-6">Full Text</a></li><li>Wei, Wu-Xiao, Zhuo, Xin-Gui, Chen, Hong-Qiao, Chen, Ming-Li (2025) <em>Status and influencing factors of balance in middle-aged and older adults with Parkinson’s disease: a national longitudinal study.</em> <strong>Front Neurol</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40371073/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12074968/">Full Text</a></li><li>Agoriwo et al. (2024) <em>System-level barriers account for non-compliance to physiotherapy among persons with Parkinson’s disease at the Korle Bu Teaching Hospital, Ghana: an explanatory mixed-method study.</em> <strong>Afr Health Sci</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40190500/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/40190500">Full Text</a></li><li>Dibble, Leland E, Ellis, Theresa D (2021) <em>The sobering and puzzling reality of rehabilitation referrals for Parkinson disease.</em> <strong>Parkinsonism Relat Disord</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33551313/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8670563">Full Text</a></li><li>Caproni, Stefano, Di Fonzo, Alessio, Colosimo, Carlo (2025) <em>Oxidative Stress: A New Pathophysiological Pathway in Parkinson’s Disease and a Potential Target of the Brain-Sport Crosstalk.</em> <strong>Parkinsons Dis</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40162062/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11952919/">Full Text</a></li><li>Aseel Aburub, Julius Sim, Sean J Ledger, Susan M Hunter (2020) <em>Cardiopulmonary Function and Aerobic Exercise in Parkinson’s: A Systematic Review of the Literature.</em> <strong>Movement disorders clinical practice</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32775504/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7396860">Full Text</a></li><li>Landers, Merrill R, Ellis, Terry D (2020) <em>A Mobile App Specifically Designed to Facilitate Exercise in Parkinson Disease: Single-Cohort Pilot Study on Feasibility, Safety, and Signal of Efficacy.</em> <strong>JMIR Mhealth Uhealth</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/33016887/">PubMed</a> | <a href="https://mhealth.jmir.org/2020/10/e18985/PDF">Full Text</a></li><li>Agley, Ledia, Hartley, Peter, Lafortune, Louise (2024) <em>Exploring the experiences, priorities and preferences of people living with Parkinson’s on exercise and physical activity promotion in the UK.</em> <strong>PLoS One</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38865427/">PubMed</a> | <a href="https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0304223&amp;type=printable">Full Text</a></li><li>Hirsch, Mark A, van Wegen, Erwin E H, Newman, Mark A, Heyn, Patricia C (2018) <em>Exercise-induced increase in brain-derived neurotrophic factor in human Parkinson’s disease: a systematic review and meta-analysis.</em> <strong>Transl Neurodegener</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/29568518/">PubMed</a> | <a href="https://doi.org/10.1186/s40035-018-0112-1">Full Text</a></li><li>Bispo, Dharah P C F, Lins, Carla C S A, Hawkes, Kelly L, Tripp, Shae, Khoo, Tien K (2024) <em>The Positive Effects of Physical Activity on Quality of Life in Parkinson’s Disease: A Systematic Review.</em> <strong>Geriatrics (Basel)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39051258/">PubMed</a> | <a href="https://hdl.handle.net/10072/431749">Full Text</a></li><li>Cheng, Yi-Chen, Su, Chun-Hsien (2020) <em>Evidence Supports PA Prescription for Parkinson’s Disease: Motor Symptoms and Non-Motor Features: A Scoping Review.</em> <strong>Int J Environ Res Public Health</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/32331349/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/7215784">Full Text</a></li><li>Xie et al. (2025) <em>Optimal dose and type of exercise improve walking velocity in adults with Parkinson’s disease: a systematic review and Bayesian network meta-analysis.</em> <strong>Sci Rep</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39824889/">PubMed</a> | <a href="https://doi.org/10.1038/s41598-025-85456-7">Full Text</a></li><li>Bosch-Barceló et al. (2025) <em>Gamification Integration in Technological Devices for Motor Rehabilitation in Parkinson Disease: Scoping Review.</em> <strong>JMIR Serious Games</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/40614259/">PubMed</a> | <a href="https://doi.org/10.2196/69433">Full Text</a></li><li>Merve Kurt, Turhan Kahraman, Zuhal Abasıyanık (2022) <em>COVID-19 and Physical Activity Behaviour in People with Neurological Diseases: A Systematic Review.</em> <strong>Journal of developmental and physical disabilities</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35125854/">PubMed</a> | <a href="https://doi.org/10.1007/s10882-022-09836-x">Full Text</a></li><li>Włodzimierz Mrówczyński (2019) <em>Health Benefits of Endurance Training: Implications of the Brain-Derived Neurotrophic Factor-A Systematic Review.</em> <strong>Neural plasticity</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/31341469/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/6613032">Full Text</a></li><li>John L Palmieri, Judith E Deutsch, Lesley Jones, Margaret Schenkman (2024) <em>Bicycling for Rehabilitation of Persons With Parkinson Disease: A Scoping Review.</em> <strong>Journal of neurologic physical therapy : JNPT</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38693613/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/38693613">Full Text</a></li><li>Macpherson et al. (2025) <em>Feasibility of a Telehealth Physical Activity Coaching Intervention for Degenerative Cerebellar Ataxia.</em> <strong>J Neurol Phys Ther</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/41171420/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/41171420">Full Text</a></li><li>Bliss, Rebecca R, Church, Frank C (2021) <em>Golf as a Physical Activity to Potentially Reduce the Risk of Falls in Older Adults with Parkinson’s Disease.</em> <strong>Sports (Basel)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/34070988/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/8224548">Full Text</a></li><li>Duranti, Elisa, Villa, Chiara (2024) <em>From Brain to Muscle: The Role of Muscle Tissue in Neurodegenerative Disorders.</em> <strong>Biology (Basel)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/39336146/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11428675">Full Text</a></li><li>Jäggi et al. (2023) <em>Feasibility and effects of cognitive-motor exergames on fall risk factors in typical and atypical Parkinson’s inpatients: a randomized controlled pilot study.</em> <strong>Eur J Med Res</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36647177/">PubMed</a> | <a href="https://eurjmedres.biomedcentral.com/counter/pdf/10.1186/s40001-022-00963-x">Full Text</a></li><li>Ren, Jimin, Patel, Neha, Johnson, Talon, Querry, Ross, Shearin, Staci (2025) <em>Skeletal Muscle &lt;sup&gt;31&lt;/sup&gt;P Magnetic Resonance Spectroscopy Study of Patients with Parkinson’s Disease: Energy Metabolism and Exercise Performance.</em> <strong>Diagnostics (Basel)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/41153248/">PubMed</a> | <a href="https://pubmed.ncbi.nlm.nih.gov/41153248">Full Text</a></li><li>E Macklin et al. (2017) <em>The impact of Tai Chi and Qigong mind-body exercises on motor and non-motor function and quality of life in Parkinson’s disease: A systematic review and meta-analysis.</em> <strong>Parkinsonism &amp; related disorders</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/28602515/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/5618798">Full Text</a></li><li>Patterson et al. (2022) <em>Study in Parkinson’s disease of exercise phase 3 (SPARX3): study protocol for a randomized controlled trial.</em> <strong>Trials</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/36203214/">PubMed</a> | <a href="https://trialsjournal.biomedcentral.com/counter/pdf/10.1186/s13063-022-06703-0">Full Text</a></li><li>Lin et al. (2024) <em>Comparing the effects of dual-task training and single-task training on walking, balance and cognitive functions in individuals with Parkinson’s disease: A systematic review.</em> <strong>Chin Med J (Engl)</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/38716704/">PubMed</a> | <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/11230837">Full Text</a></li><li>Barbosa et al. (2022) <em>How I treat Parkinson’s disease.</em> <strong>Arq Neuropsiquiatr</strong> — <a href="https://pubmed.ncbi.nlm.nih.gov/35976316/">PubMed</a> | <a href="https://www.scielo.br/j/anp/a/GYfpYcFSj4GrYNCLccDZQrx/?lang=en&amp;format=pdf">Full Text</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dd719da16ef0" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>