<?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 Daria Meshcherina on Medium]]></title>
        <description><![CDATA[Stories by Daria Meshcherina on Medium]]></description>
        <link>https://medium.com/@dmescherina?source=rss-6471f9151c6b------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*CJqTcmiQVEcTaPcGAqLh1Q.jpeg</url>
            <title>Stories by Daria Meshcherina on Medium</title>
            <link>https://medium.com/@dmescherina?source=rss-6471f9151c6b------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 16:09:22 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dmescherina/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[How To Be Full(ish) on a Calorie Deficit: My Data-Driven Survival Guide]]></title>
            <link>https://medium.com/in-fitness-and-in-health/how-to-be-full-ish-on-a-calorie-deficit-my-data-driven-survival-guide-3146de9a2f5f?source=rss-6471f9151c6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/3146de9a2f5f</guid>
            <category><![CDATA[fitness]]></category>
            <category><![CDATA[nutrition]]></category>
            <category><![CDATA[health]]></category>
            <category><![CDATA[weight-loss]]></category>
            <category><![CDATA[wellness]]></category>
            <dc:creator><![CDATA[Daria Meshcherina]]></dc:creator>
            <pubDate>Wed, 29 Apr 2026 06:11:52 GMT</pubDate>
            <atom:updated>2026-04-29T06:11:52.738Z</atom:updated>
            <content:encoded><![CDATA[<h4>What I do when the calorie budget is smaller than my appetite</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kC_1E02H72OB6HJV" /><figcaption>Photo by <a href="https://unsplash.com/@limamauro23?utm_source=medium&amp;utm_medium=referral">Mauro Lima</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The sad but unavoidable truth of weight loss is that you have to create a calorie deficit. You eat less than you burn, you lose weight. Simple, elegant, annoyingly non-negotiable. The problem isn’t understanding the concept. The problem is living inside it, day after day, when your body hasn’t quite got the memo that the calorie tracking app says you’re done eating.</p><p>I know the moment well. It’s mid-afternoon, or just after dinner, or honestly sometimes both — and I’m staring at my remaining calorie budget like maybe if I look at it long enough, the numbers will change. They don’t. I have, let’s say, 300 calories left and the kind of hunger that feels deeply unreasonable given how much I’ve already eaten. This isn’t a willpower problem. This is just what a calorie deficit feels like from the inside.</p><p>For a while, my strategy was essentially just suffering through it. Smaller portions of the same foods, slightly more heroic willpower, roughly the same results. <br>Then I tried the high-volume approach — eating enormous bowls of lettuce and feeling virtuous and hungry simultaneously. Then the high-protein approach, which worked better but still left a specific kind of evening hunger that no amount of chicken breast could solve. <br>What none of these approaches had was a framework. I was pulling levers at random.</p><p>So I did what I do when something isn’t working: I started treating it as a data problem.</p><p>So what do you do when hunger is non-negotiable, and calories are finite? You optimise for <strong>satiety per calorie</strong>. Which means asking a question: <em>“What will actually make me feel full?”</em></p><p>Satiety is doing a lot of heavy lifting that most diet advice glosses over. Three factors matter most: <a href="https://www.sciencedirect.com/science/article/abs/pii/S0268005X1630340X"><strong>protein</strong></a><strong>, </strong><a href="https://www.sciencedirect.com/science/article/abs/pii/S0023643821007398"><strong>fibre</strong></a><strong>, and </strong><a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC4182946/"><strong>food volume</strong></a><strong> </strong>(or more precisely, low energy density). Those are the three levers to use when trying to get maximum satiety bang for your calorie buck.</p><p>Below are the foods that actually deliver.</p><h3>Protein Heroes</h3><p>Protein is the most effective macronutrient for satiety — it triggers hormonal responses, raises blood amino acid levels, and even slightly increases energy expenditure. In plain terms, it keeps you full longer per calorie than either fat or carbohydrate. The list skews heavily towards fish, dairy, and lean meat, but note that firm tofu and edamame hold their own if you’re plant-based.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iSkoJi6kYAa7s3bk10GyIg.png" /><figcaption>Protein per 100 kcal — higher is better. Chart by the author.</figcaption></figure><p>What struck me most when I ran these numbers was how much the dairy entries hold up. Quark and Greek yoghurt sitting above chicken breast wasn’t something I expected — but at 19.4g and 17.5g of protein per 100 kcal respectively, they’re genuinely competitive, and they come with the added advantage of being cold, fast, and requiring zero cooking at 4 pm.</p><p>The other thing worth noting: the gap between the top three and the bottom three is enormous. A 100-calorie portion of chickpeas gives you roughly a quarter of the satiety protein hit of the same calories in tuna. Not a reason to avoid chickpeas — but a reason to pair them with something higher up the list.</p><h3>Fiber Heroes</h3><p>Fibre adds bulk and weight to food without adding proportional calories — and viscous fibre specifically slows gastric emptying, meaning your stomach stays physically fuller for longer. Chia seeds are the outlier at the top (soaked, not dry — the water absorption is the whole point). Everything else is vegetables and legumes doing quiet, unglamorous work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YQ9B1ZiitpYCgv7HEjCr2A.png" /><figcaption>Fibre per 100 kcal — higher is better. Chart by the author.</figcaption></figure><p>The chia seed number is genuinely startling until you understand the mechanism — soaked chia seeds are mostly water, and that water is bound inside a gel structure that expands in your stomach.</p><p>The more interesting observation for me is how many of the fibre heroes are also the volume heroes: spinach, broccoli, cauliflower and strawberries all appear in both lists. That overlap is not a coincidence — high-fibre foods tend to be high-water foods, which is exactly what drives low energy density. The two levers are, in many cases, the same lever.</p><h3>Volume Heroes</h3><p>Food volume works mechanically: satiety is first triggered by your gut’s physical distension response — how much is actually in there — rather than by caloric content. The numbers here are striking. You can eat 714g of celery for 100 calories. Your stomach has no idea that’s “not many calories.” It just registers: full.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1CNm8zpvWTyYytMYC-twFw.png" /><figcaption>Grams of food per 100 kcal — higher is better. Chart by the author.</figcaption></figure><p>This chart is the one that changed how I think about building a plate. A former colleague of mine used to insist, with genuine conviction, that a cucumber shouldn’t be categorised as food. “It’s just crunchy water,” he’d say. And honestly , on a calorie deficit, crunchy water that fills your stomach for 16 calories per 100g is exactly what you want.</p><p>The foods at the top — celery, lettuce, cucumber, courgette — aren’t particularly exciting on their own. But as a base, a bed, a bulk-builder, they’re extraordinary. 150g of spinach is essentially nothing calorically (around 35 kcal), but it physically fills a large bowl, takes time to eat, and occupies real estate in your stomach. The reframe that helped me most: stop thinking of these foods as sides or garnishes. They are the bearing walls of the plate. Everything else — the protein, the more calorie-dense elements — goes on top of them.</p><h3><strong>Honourable Mentions</strong></h3><p>Numbers are great until they aren’t. A top-15-per-metric ranking is a clean way to think about satiety, but it’s also a blunt instrument — and a few foods fall through the gaps precisely because their value doesn’t sit neatly in a single column. These didn’t make any chart. They should have been on all of them.</p><h4>Nuts</h4><p>Nuts seem like a strange inclusion given their calorie density, but there’s a genuinely interesting mechanism at work: a meaningful portion of the fat in nuts is trapped inside fibrous cell walls that your gut never fully breaks down. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC10334117/">Studies suggest</a> you absorb around 20–30% fewer calories from whole nuts than the label implies. They’re also protein and fibre combined, they suppress appetite hormonally, and they’re one of the few genuinely portable, no-prep satiety foods. Just don’t eat them straight from the bag while watching television.</p><p>Ask me how I know.</p><h4>Potatoes</h4><p>Potatoes are the biggest anomaly on this list.</p><p>They didn’t make the volume chart — at around 115g per 100 kcal, they’re respectable but not exceptional. What makes them remarkable is <a href="https://pubmed.ncbi.nlm.nih.gov/7498104/">Holt’s 1995 Satiety Index</a>, where boiled potatoes scored 323 — nearly three times higher than white bread and higher than anything else tested.</p><p>The mechanism goes beyond volume: <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC7049780/">potatoes trigger strong hormonal satiety signals</a>. Cook them, then cool them. That’s when resistant starch forms — a type of fibre your small intestine can’t digest, which feeds gut bacteria and blunts blood sugar response. Cold potato salad, for the record, qualifies.</p><h4>Eggs</h4><p>Whole eggs missed the protein chart because the yolk fat pulls the protein-per-calorie ratio down. But the yolk is precisely what makes eggs so satiating — it triggers a <a href="https://pubmed.ncbi.nlm.nih.gov/8773735/">particularly strong release of cholecystokinin</a>, a satiety hormone that signals fullness to the brain more powerfully than almost any other food component. An egg is also one of the most complete, fast, cheap, and flexible foods available on a deficit.</p><h4>Broth</h4><p>Bone broth or protein-rich broth is doing at least two things simultaneously: it’s almost pure protein in liquid form, which hits the hormonal satiety pathway, and it’s also high volume by definition — a large bowl of broth is physically filling in a way a protein bar with the same macros simply isn’t. <br>There’s also something worth acknowledging that isn’t in any of the studies: a warm bowl of broth on a cold evening also warms up your soul. That matters more than nutritionists tend to admit.</p><h4>Legumes</h4><p>Whole legumes broadly — lentils and chickpeas made the charts, but the category earns a collective mention: protein, fibre, and resistant starch in a single food, which means all three hero metrics plus the potato satiety effect simultaneously. If you’re plant-based and the protein chart felt discouraging, build around this.</p><h4>Popcorn</h4><p>Air-popped popcorn narrowly missed the volume chart — at 380g per 100 kcal, it’s remarkable — and it takes a genuinely long time to eat, which matters. The only snack that functions as a bearing wall. Just not the cinema kind.</p><h4>Konjac</h4><p>Shirataki or konjac noodles<strong> </strong>have almost zero calories, pure glucomannan fibre that gels in your stomach and delays gastric emptying significantly. The texture is a separate conversation. But if you make peace with it, you’ve found a cheat code.</p><h3>Practical Takeaways</h3><h4>The overachievers</h4><p>Some foods don’t just show up on one chart. They show up on all of them.</p><p>Spinach lands on both the fibre list (9.6g per 100 kcal) and the volume list (435g of food per 100 kcal). So do broccoli, cauliflower, and strawberries. Greek yoghurt bridges protein and volume. Lentils span protein and fibre. These are the foods that pull multiple satiety levers at once — which means you’re not choosing between strategies, you’re stacking them. If I had to build a shortlist of foods to always have in the fridge on a deficit, it would start here.</p><h4>My actual fridge on a deficit</h4><p>My go-to satiety staples are: 0% Greek yoghurt (bridges protein and volume, endlessly versatile), spinach (volume and fibre, wilts into nothing when cooked), canned tuna (cheapest protein per calorie I’ve found), a jar of chia seed and thawed frozen raspberry jam that I make in about four minutes and put on everything, and pears — my default evening fruit, quiet and reliable and surprisingly good on the fibre front.</p><p>For breakfast, the same chia seeds go into a pudding with protein powder: fibre, protein, and zero decisions before 8 am. All three levers covered, enough combinations to not lose my mind by mid-week.</p><h4>The 4 pm scenario</h4><p>When calorie tracker and stomach are at an impasse it calls for all three heroes at once. My go-to is 150g of 0% Greek yoghurt with 80g of raspberries and a tablespoon of soaked chia seeds. That’s roughly 200 calories, but it’s 18g of protein, 10.6g of fibre, and 305g of food. All three heroes, one bowl.</p><h4>The still-hungry-after-dinner scenario</h4><p>This situation calls for a different approach — one that leans harder on volume. A plate of 150g spinach, 150g cucumber, 150g tomatoes and 100g of canned tuna comes in at around 186 calories, 29g of protein, and 550g of actual physical food on your plate. That is not a small amount of food. Your stomach doesn’t know it’s 186 calories. It just knows it’s full.</p><p>None of this is a magic fix. A calorie deficit is still a calorie deficit, and some days hunger is just going to win. But there’s a meaningful difference between hunger you’re managing and hunger you’re just enduring. Before I had this framework, every deficit felt like deprivation — like something was being taken away. Now it feels more like a resource allocation problem, which is, frankly, a much more comfortable thing to be bad at. Knowing which foods give me the best return on my calorie investment hasn’t made the deficit disappear.</p><p>It just made it livable.</p><p>And there is something to be said for opening the fridge without the anxiety.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3146de9a2f5f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/in-fitness-and-in-health/how-to-be-full-ish-on-a-calorie-deficit-my-data-driven-survival-guide-3146de9a2f5f">How To Be Full(ish) on a Calorie Deficit: My Data-Driven Survival Guide</a> was originally published in <a href="https://medium.com/in-fitness-and-in-health">Second Nature (formerly In Fitness &amp; In Health)</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[In Defence of a Wandering Mind]]></title>
            <link>https://medium.com/in-fitness-and-in-health/in-defence-of-a-wandering-mind-b0934a21454e?source=rss-6471f9151c6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/b0934a21454e</guid>
            <category><![CDATA[health]]></category>
            <category><![CDATA[mental-health]]></category>
            <category><![CDATA[neuroscience]]></category>
            <category><![CDATA[wellness]]></category>
            <category><![CDATA[poetry]]></category>
            <dc:creator><![CDATA[Daria Meshcherina]]></dc:creator>
            <pubDate>Wed, 04 Mar 2026 10:46:00 GMT</pubDate>
            <atom:updated>2026-03-04T10:46:00.800Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HcmTcIUv0Oua_OBm" /><figcaption>Photo by <a href="https://unsplash.com/@matteo___c?utm_source=medium&amp;utm_medium=referral">Matteo Costa</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h4>How poetry helps you be more creative</h4><p>You’ve been told to focus. To concentrate. To stay on task.</p><p>But what if the most productive thing your brain can do mid-afternoon is quietly drift away — ideally via a Kipling poem?</p><p>Mind wandering has a reputation problem. We call it distraction, daydreaming, not paying attention. And in the era engineered to hijack your attention — notifications, autoplay, the algorithmic infinite scroll — that reputation has only gotten worse.</p><p>The modern attention economy treats your focus as a resource to be extracted; every app, every platform is competing for the same finite slice of your cognitive bandwidth.</p><p>Slow, unstructured thought isn’t just undervalued. It’s under attack.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZD_Xd0A7lA3irESZ2CdAYw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@ucaremre35?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Emre Ucar</a> on <a href="https://unsplash.com/photos/blurred-figures-in-a-crowd-at-night-kw7rYsVhqcM?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>This isn’t just a modern problem.</p><p>During the first Covid lockdown, I found myself pulling Bertrand Russell’s essays off the shelf — the kind of reading you only do when the diary clears, and the ambient noise of obligation finally goes quiet. I came across <a href="https://harpers.org/archive/1932/10/in-praise-of-idleness/">an essay from 1932</a> where he wrote that <em>“there was formerly a capacity for light-heartedness and play which has been to some extent inhibited by the cult of efficiency”</em> — and that <em>“the modern man thinks that everything ought to be done for the sake of something else, and never for its own sake.”</em></p><p>Russell was writing about industrial labour. He might as well have been describing my Tuesday afternoon inbox. Nearly a century later, the cult of efficiency hasn’t loosened its grip. If anything, it found new means to reach a larger part of our lives.</p><p>The slower life — the one with margin in it, with unhurried thought and unscheduled afternoons — is rarely accidental. You have to carve it out deliberately, and then defend it.</p><p>Ask me how well I manage to do this.</p><p>At the same time, the scientific evidence for why it’s worth the fight keeps accumulating. One of the most compelling benefits of idleness is also one of the least obvious: letting your mind roam is not the enemy of creative thinking. It may be a prerequisite for it.</p><blockquote>Two roads diverged in a yellow wood,<br>And sorry I could not travel both<br>And be one traveler, long I stood<br>And looked down one as far as I could<br>To where it bent in the undergrowth;</blockquote><p><em>— Robert Frost</em></p><p>The data, it turns out, is on Russell’s side. This <a href="https://research.gold.ac.uk/id/eprint/38386/10/Poetic%20Break%20Incubation%20for%20Associative%20Creativity.pdf">new study</a> by Chaudhuri, S. &amp; Bhattacharya, J. (2025) tested this idea and gave mind wandering a poetic ally.</p><p>Participants were asked to take a break from a creative task.</p><p>Some got poetry. Some got poetry with an analytical task attached. Some got prose.</p><p>The poetry readers came back with measurably better creative flow — not wilder ideas, but ideas that <em>connected</em>. There’s a difference, and it matters.</p><h4>The Study Question</h4><p>The researchers were interested in a well-established phenomenon called <em>incubation</em> — the idea that stepping away from a problem, rather than grinding harder at it, can unlock creative thinking.</p><p><a href="https://pubmed.ncbi.nlm.nih.gov/17073528/">Prior</a> <a href="https://pubmed.ncbi.nlm.nih.gov/25293689/">research</a> <a href="https://pubmed.ncbi.nlm.nih.gov/22941876/">links</a> this effect to mind wandering. Chaudhuri and Bhattacharya wanted to know whether <em>reading poetry specifically could act as an effective incubation trigger</em> — whether poetic text, with its capacity for self-reflection and emotional resonance, might prime the mind for creative thought more effectively than ordinary text.</p><h4>The Study Design</h4><p>153 participants completed creative tasks both before and after a reading break. They were divided into three groups: one read Rudyard Kipling’s “<em>If</em>”, another read and rated the poem, and a third read a non-poetic text of comparable length. Participants also reported their levels of mind wandering during the break, and completed a questionnaire measuring their general daydreaming tendencies.</p><p>Creativity was assessed in two distinct ways — <em>Forward Flow</em> (FF), which measures associative thinking, the ability to move fluidly from one idea to the next; and the <em>Alternate Uses Task</em> (AUT), which measures divergent thinking, the capacity to generate novel, out-of-the-box ideas.</p><h4>The Study Results</h4><p>High mind-wandering in the poetry-reading group significantly boosted associative thinking. Low mind-wandering in that same group actually <em>decreased</em> it.</p><p>In contrast, associative thinking improved in the other conditions regardless of mind-wandering levels. No significant effects on divergent thinking were observed across conditions, and daydreaming traits did not influence the results.</p><p>In other words: poetry didn’t make people wilder or more unconventional thinkers. But it did make their ideas flow more freely — one thought leading naturally to the next. And crucially, this <em>only held within the poetry-reading group when minds actually wandered</em>. The other groups — those rating the poem or reading prose — showed improvement regardless.</p><h4>The Nuance That Matters</h4><p>The distinction between associative thinking and divergent thinking is easy to gloss over, but it matters.</p><p><em>Divergent thinking is the spark</em> — the capacity for a genuinely novel idea.<em> Associative thinking is the current</em> — the<em> </em>ability to navigate between concepts, to build chains of thought, to follow an idea to somewhere unexpected but logical. Most creative work needs both, but the one that most often stalls mid-project, mid-afternoon, mid-draft, is the latter.</p><p>And that’s exactly where poetry appears to help.</p><h4>Takeaways</h4><blockquote>Let us, then, be up and doing, <br>With a heart for any fate; <br>Still achieving, still pursuing, <br>Learn to labour and to wait.</blockquote><p><em>— Henry Wadsworth Longfellow</em></p><p>We live in an age that worships directed attention — the focused mind, the optimised workflow, the productivity system that leaves no idle moment unaccounted for.</p><p>And yet.</p><p>The evidence keeps accumulating, quietly and rather stubbornly, that the wandering mind is not the enemy of good thinking but its secret collaborator. To let go of the task, to stop <em>trying</em>, to allow thought to drift without destination — this is not laziness. It is, increasingly, what the data and neuroscience recommend.</p><p>Even writing this article, I found myself reaching for poems at odd moments — not as research, but as relief. Small detours that turned out to be shortcuts.</p><p>Poetry turns out to be a remarkably efficient door into a creative state: a few lines, a little strangeness, and the mind slips its leash almost without noticing.</p><p>Poetry, however, is not a creative detonator. It won’t blow the doors off your thinking. But, as the author of the paper put it herself: “<em>Poetry offers a unique and subtle way to allow the mind to wander and uncover surprising connections within the landscape of our thoughts… So, the next time you take a break, consider dipping into poetry — it might just lead you to unexpected creative insights.</em>”</p><p>We’ve spent decades trying to improve our attention when we should have been learning to lose it on purpose.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jKKpZ7j_wrP5gUjvHL9yeQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@marekpiwnicki?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Marek Piwnicki</a> on <a href="https://unsplash.com/photos/a-hammock-hanging-between-two-trees-on-a-beach-WkbosUDx9uQ?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b0934a21454e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/in-fitness-and-in-health/in-defence-of-a-wandering-mind-b0934a21454e">In Defence of a Wandering Mind</a> was originally published in <a href="https://medium.com/in-fitness-and-in-health">Second Nature (formerly In Fitness &amp; In Health)</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Fertility Numbers Few Talk About]]></title>
            <link>https://medium.com/in-fitness-and-in-health/the-fertility-numbers-few-talk-about-d2873cb267fc?source=rss-6471f9151c6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/d2873cb267fc</guid>
            <category><![CDATA[fertility]]></category>
            <category><![CDATA[wellness]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[health]]></category>
            <category><![CDATA[fertility-treatment]]></category>
            <dc:creator><![CDATA[Daria Meshcherina]]></dc:creator>
            <pubDate>Mon, 16 Feb 2026 06:36:43 GMT</pubDate>
            <atom:updated>2026-02-16T06:36:43.153Z</atom:updated>
            <content:encoded><![CDATA[<h4>What I learned pulling apart IVF statistics so you don’t have to</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kXG6qRabQtBBsS5Z3CRe8A.png" /><figcaption>Photo by Daniel Jericó on Unsplash</figcaption></figure><p>Do you, like me, have this vague memory from high school sex ed where the implicit conclusion was that spending too much time in proximity to a girl — let alone touching her hand — meant she’d have a high chance of immediately getting pregnant?</p><p>This perception, of course, is spectacularly far from reality. The fertility journey is uncertain, variable, and often bewilderingly difficult.</p><p>I only started becoming aware of actual fertility data a few years ago, when I was going through family planning with my partner and, later, an IVF round. Despite being reasonably clued-in about health and medicine, I found much of the data surprising, some of it overwhelming, and nearly all of it buried under layers of jargon that seemed specifically designed to confuse anyone trying to make an informed decision.</p><p>But as always, when faced with impenetrable data, I decided not to give up, but to systematically pull it apart instead.</p><p><strong>This article is my attempt to do what I wish someone had done for me earlier: distil the key statistics around fertility and IVF into something genuinely digestible.</strong> I want to give you an at-a-glance overview that doesn’t require a statistics degree — but that’s still packed with the eye-opening numbers I wish I’d known sooner.</p><p>Whether you’re actively going through the process, curious about your options, or simply wondering what this landscape actually looks like beneath the clinic marketing speak, I hope that you’ll walk away with a clearer picture.</p><p>There’s unfortunately a lot of misleading information out there — and not just on dodgy forums. A <a href="https://www.bbc.co.uk/news/uk-68505321">2024 BBC investigation</a> found that <strong>41% of UK clinics offering egg freezing could be breaching advertising guidance</strong> by failing to make clear a patient’s actual chances of having a baby. Most were advertising “successful thaw rates” of 80–95% — which sounds reassuring until you realise that’s just one early step in a multi-stage process.</p><p>The actual live birth rate? More like 20–30% per round for IVF, and as low as 5% for patients in their 40s. As one fertility charity put it: patients are being “misled by this level of information.”</p><p>And the problem isn’t limited to egg freezing. When I tried to piece together a coherent picture from individual IVF clinic websites, I quickly realised there was no single standard for what they had to disclose. Different ratios, different definitions, no clarity on how many women were included in each number or what their circumstances were.</p><p>A <a href="https://bmjopen.bmj.com/content/7/1/e012218">University of Manchester study</a> found clinics use <strong>33 different ways</strong> to report pregnancy rates alone, meaning even thorough prospective patients end up comparing apples to oranges. Some report pregnancy rates; others report live births. Some cherry-pick their best patient cohorts. Some use “per embryo transfer” instead of “per embryo transferred” — a subtle distinction that can make the same clinic look dramatically different.</p><p>So I turned to the UK regulator data and started from there.</p><p>I’ll do my best to share what I learned, but please consult a qualified medical professional for advice relevant to your specific situation. This is a data-nerd’s map of the landscape, not medical guidance. And most importantly, as is the case with statistics, it looks at averages, which can be very different from your individual circumstances.</p><h3>Fertility vs. Time</h3><p>As much as I personally object to the term “geriatric mother” applied to women after 35 (my partner was affectionately referring to me as “Gerry” for a while as a result), there is no denying that fertility decreases with age.</p><p>A <a href="https://www.science.org/doi/10.1126/science.3755843">landmark study from 1986</a> confirmed that fertility declines particularly quickly from mid-30s. <a href="https://pubmed.ncbi.nlm.nih.gov/28917614/">Modern research</a> also confirms this: <strong>a healthy, fertile 30-year-old woman has about a 20% chance, or 1-in-5, of getting pregnant each cycle. By age 40, this drops to less than 5%, or 1-in-20, per cycle.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eIEkGb8WmgwExXQ1Tjyquw.png" /><figcaption>Probability of Natural Conception by Age. Schematic representation showing key inflection points at ~32 and ~37 years. Based on Menken et al. (1986), Wesselink et al. (2017), Hutterite population data, and related studies.</figcaption></figure><p>To interpret these probabilities, you have to remember that these numbers represent expected outcomes in a large number of repetitions.</p><p>For example, if you are trying for a baby with chances 1-in-5, then on average over thousands of months, you would get pregnant every 5 months. If, however, your chances are 1-in-20, you would, on average, get pregnant once every 1.7 years.</p><p>Bear in mind, these probabilities also do not increase if you’re trying consecutive months — every month, your chances stay the same. This helps to put into perspective just how meaningful these changes in fertility are for someone looking to conceive — and they happen over a span of just 5 years. It’s also tremendously important to note that the ages of 35 and 40 here are indicative — and for some women, these changes can happen earlier or later, and their chances will also depend on how high or low their original fertility to start with.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NrX9jdc55o34ZVWvh9uT-w.png" /></figure><h3>Quantity vs. Quality</h3><p>A lot of the measures of fertility, like AMH hormone level or follicular reserve, look at the quantity of available reproductive stock in the body. However, this is only part of the story. There is a long journey from egg and sperm cells to a healthy embryo to live birth, and I personally found the conversion rates along the way quite startling.</p><p>One of the main determinants of how many eggs become a healthy embryo is cell quality, not the quantity, and this cell quality is also declining with age.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nfymTwC7juBuyVINSZuLAw.png" /><figcaption><strong>Average outcomes of one IVF cycle per age group based on clinical literature.</strong> Conversion rates: ~70% fertilisation, ~40% blastocyst survival, genetic health varies by age (55% under 35, 40% at 35–40, 25% over 40), ~55% live birth per genetically healthy embryo. Live birth values below 1.0 indicate multiple cycles are typically needed for one baby.</figcaption></figure><p>Let’s say we start with 20 eggs extracted during an IVF cycle, which is quite a remarkable number already. For reference, clinical data shows that the average number of eggs per extraction is 10-20 for women under 35, 6-12 eggs for women between 35 and 40, and 4-6 for women over 40 — yet another influence of age.</p><p>Roughly two‑thirds to three‑quarters of mature eggs will successfully fertilise — around 12–15 out of 20 in our example, depending on age, sperm quality, and lab conditions.</p><p>Over the next 3–5 days, as fertilised eggs grow into blastocysts, only about 30–40% will survive to that stage — in this example, around 5–7 embryos.</p><p>Now this is where genetic cell quality comes into play. As we get older, the natural rate of genetic errors in the reproductive cells increases. For a woman around 35, roughly half of those blastocysts will have the correct number of chromosomes. In our example of six blastocysts, that’s on the order of three, perhaps four, to have the correct number of chromosomes and thus be viable to become a healthy baby. By around 40, only about a third of blastocysts are chromosomally normal — roughly two out of six. As you move into the early 40s, that often falls closer to one or two out of six.</p><p>Given every genetically healthy embryo has a 45-65% chance to become a live healthy baby (and this proportion is less variable with age), we need<strong> 2–3 genetically healthy embryos to approach a 95% probability of a live birth</strong> over potentially multiple transfers.</p><p>Working back from these numbers to imply how many eggs, and hence IVF cycles, one might need to achieve a live birth can be quite daunting. Given physical, emotional, and financial investments in the process, it is only natural for one to want some guarantees around the quality and success of the outcomes from the chosen clinics. Which is why I thought it was important to dig through the incoherent metrics advertised and do a clean comparison of fertility clinics. Armed with our understanding of fertility data, its variability by age, and different stages of the process, let’s delve into the strategy for clinic comparison.</p><h3>The Data Strategy</h3><p>Given all the uncertainty and variability in fertility outcomes, how can someone measure the performance of the clinics and compare them?</p><p>Having already learned how inconsistently clinics report their data, I turned to the UK regulator —<a href="https://www.hfea.gov.uk/"> Human Fertilisation &amp; Embryology Authority</a>, HFEA.</p><p><a href="https://www.hfea.gov.uk/choose-a-clinic/">HFEA&#39;s “Choose A Clinic” page</a> contains a very useful dataset of statistics from multiple fertility clinics between 1991 and 2023, albeit sparse in many cases. What I found particularly valuable and reassuring was that the data is validated and metrics are computed from common denominators, and clean assignment to age cohorts is performed.</p><p>What I did was get the list of 5 clinics in my area from a Google search, and then look at the HFEA statistics for each. I then collated them all in one table, compared the metrics, and ranked the clinics accordingly. I then weighted the ranking in each category according to what I deemed was most important, to arrive at the overall ranking. Below are the data points I paid attention to, some that I constructed myself, and why I found them useful.</p><p>Some overarching principles:</p><ul><li>I have looked at all the data for my specific age group</li><li>It matters less what type of success metric you choose specifically (per embryo transferred, per IVF cycle, etc.), as long as you consistently extract the same one across all the clinics you are comparing</li></ul><ol><li>IVF pregnancy rates/live birth rates from fresh IVF cycles. These, to me, most straightforwardly indicate success rates of the clinic’s IVF protocols.</li><li>IVF pregnancy rates/live birth rates from the frozen IVF cycles. Because we planned to freeze the embryos, this statistic also directly mattered to me.</li><li><strong>A difference between frozen and fresh pregnancy rates/live birth rates.</strong> With modern technologies, the thawing success rate should be close to 98–99%, so a substantial gap in either direction likely reflects something else — patient selection practices, embryo grading criteria, transfer timing protocols, or lab-specific factors. I couldn’t tell you exactly what a gap means, but I can tell you it’s worth asking about.</li><li>Given that knowing if your embryos are genetically healthy can bump the live birth rate from around 30% to 60%, it was important for me that the clinic provided <strong>PGT-A testing</strong> (<em>Pre-implantation Genetic Testing for Aneuploidy — essentially checking embryos have the right number of chromosomes before transfer</em>) that can test this. Not all clinics do that, and very few recommend it without the obvious genetic risk factors. So having a PGT-A mentioned as an option on the website was a good indication of openness and availability of this check.</li><li>I was very conscious of potential overstimulation and side effects of hormones, so I searched the websites for mentions of how many scans to expect during the IVF protocol, and how frequently the blood checks, particularly by the end of the cycle. The more the better.</li><li><strong>Do not dismiss the patient rating.</strong> It was very non-uniform across the clinics I looked at. Incidentally, having gone with the clinic that had a perfect patient score, I can definitely recommend it and have seen how much of a difference it makes when day-to-day operations are well organised, and you don’t have to stress about delays, scheduling issues, or second-guessing if the amount of attention and quality of care you were given were sufficient.</li><li><strong>Does the website mention specific strategies for checking if transfers are not successful?</strong> Follow-ups post implantations with a specific focus on figuring out what might have gone wrong if a few attempts are unsuccessful seems like an obvious step, but it was not always mentioned by the clinic.</li></ol><p>In the end, I had a table that looked something like this — and the choice of clinic became much more obvious:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Aq-gDyVxPZYkqcxz-pKnTw.png" /><figcaption><strong>Weighing all the factors to inform clinic choice.</strong> Example comparison table using HFEA data and clinic website research. Weights reflect my personal priorities — yours may differ. Green = best in category, red = lowest. The “Overall Score” combines weighted ranks across all factors. Raw data from HFEA “Choose a Clinic”; supplementary info (PGT-A availability, monitoring protocols, implantation follow-up) gathered from clinic websites.</figcaption></figure><p>So if you’re navigating this fertility-IVF journey yourself, I would encourage you to check out the HFEA’s “Choose a Clinic” page and build your own version of this table, or at least check the stats of your chosen clinic. Weigh up different factors in the proportion that matters to you. The data is finally there — you just have to demand it speaks plainly.</p><p>Knowing all these numbers didn’t make the IVF process easy — if anything, it became even more obvious what odds every woman and couple are going against when they start the process. I have tremendous respect for all of us facing this daunting process. But knowing more made me feel like I was finally working with real information instead of clinic-curated hope. You can’t control (all) the outcome — but you can refuse to be confused by it.</p><p>Turns out, getting pregnant is nothing like high school made it sound.</p><p>But here’s what I’ll tell my future kid when it’s their turn for that awkward class: the real lesson is that when the data looks confusing, someone usually wants it that way — and your job is to see through it anyway.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d2873cb267fc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/in-fitness-and-in-health/the-fertility-numbers-few-talk-about-d2873cb267fc">The Fertility Numbers Few Talk About</a> was originally published in <a href="https://medium.com/in-fitness-and-in-health">Second Nature (formerly In Fitness &amp; In Health)</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Lost Weight for Good with Data Science: Part 1, The Mindset]]></title>
            <link>https://medium.com/in-fitness-and-in-health/how-i-lost-weight-for-good-with-data-science-part-1-the-mindset-e70470fc2a2b?source=rss-6471f9151c6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/e70470fc2a2b</guid>
            <category><![CDATA[fat-loss]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[wellness]]></category>
            <category><![CDATA[health]]></category>
            <category><![CDATA[weight-loss]]></category>
            <dc:creator><![CDATA[Daria Meshcherina]]></dc:creator>
            <pubDate>Tue, 20 Jan 2026 21:29:37 GMT</pubDate>
            <atom:updated>2026-01-22T17:25:54.937Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZsAD1OOn622kBlwrXkBWhw.png" /></figure><h4>Why I stopped relying on willpower and started using trend lines</h4><p>Like a lot of people in the UK, I occasionally eat more than I actually require for the day — a habit that modern abundance has made remarkably easy to sustain. I’ve always been conscious of leading a healthy lifestyle and wanted to look good as an added bonus. I exercised regularly, monitored what I was eating (sometimes with obsessive precision), but despite these efforts, my weight began its inexorable upward march in my mid-20s and continued with accelerating pace.</p><p>Occasionally, I would declare war on this trend and start winning — though at what cost! Too many restrictions, too many rules, too much pressure on myself. I’d disrupt the pattern temporarily, but would inevitably return to the same upward trajectory eventually — because none of my interventions were sustainable long term. So a subtle but unavoidable weight trend — a few kilos per year — continued in the undesired direction.</p><p>One day, however, I discovered John Walker’s <a href="https://www.fourmilab.ch/hackdiet/www/">“The Hacker’s Diet”</a> and realised that what I needed wasn’t a military operation, but a fundamental debugging exercise. I required a new understanding, better tools, and — most critically — some patience. Always a data scientist, even outside work, I adapted Walker’s framework into a Python and Jupyter Notebook-friendly implementation. But you don’t need to be a Data Scientist to do the same — a lot of this approach is about your attitude, and some simple maths that a spreadsheet would be just as suitable for.</p><h3>The Mindset</h3><h4>Humans as Water Balloons with Opinions</h4><p>Walker’s first genius insight is very refreshing: humans are essentially sophisticated water containers. We’re roughly 60% water, and that water weight fluctuates wildly based on sodium intake, sleep, stress, and general levels of entropy in your surroundings. This means daily weight measurements contain more noise than signal — something I should have recognised as a data scientist, yet totally forgot when standing on bathroom scales.</p><p>The implication is liberating: that 1.5kg overnight weight gain wasn’t your body instantly converting dinner into permanent fat. It was probably just water retention having a dramatic day. Understanding this removed most of the psychological torture from my daily weigh-ins.</p><h4>The Broken Thermostat Problem</h4><p>Walker’s central idea is elegant: our appetite should work like a thermostat, automatically adjusting intake to maintain steady weight.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2s8ToxIvEtNtTXahgx10Tg.png" /><figcaption>Our hunger thermostat is an example of a system with negative feedback (adapted from Walker, J. “The Hacker’s Diet”)</figcaption></figure><p>This appetite thermostat should put you into the optimal middle bit of how many calories you eat each day.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VqtK6np6CeQgyZJbIewSVQ.png" /><figcaption>Ideally your hunger thermostat will point you to the horizontal part of this curve where you don’t systematically lose or gain weight (adapted from Walker, J. “The Hacker’s Diet”)</figcaption></figure><p>For most of human history, this system worked adequately. Food scarcity provided natural constraints, and our regulatory mechanisms evolved accordingly.</p><p>Modern abundance broke everything. We’re running Stone Age appetite software in a contemporary food environment, and the results are predictably chaotic. The system that kept us alive during famines now struggles with constant caloric availability.</p><h4>Three Types of Weight Control Systems</h4><p>Walker identifies three patterns of weight regulation, two of which represent a different type of system malfunction:</p><p><strong>Stable Sam</strong> has the perfect thermostat — intake automatically adjusts to maintain weight. These people can eat intuitively without conscious effort. If you know one, try not to resent them too obviously.</p><p><strong>Oscillating Oscar</strong> experiences dramatic weight swings. Loses (sometimes significant) amounts through restriction, then regains it all when restriction becomes unsustainable. The classic yo-yo pattern that makes any attempts at weight management feel so futile. Yours truly sits firmly in this category.</p><p><strong>Blown-up Buster</strong> has no feedback system whatsoever. Weight increases steadily over time because the thermostat has given up entirely. This is the slow, relentless drift of constant weight gain.</p><p>So here I was, an Oscillating Oscar, cycling through restriction and release without being able to settle in the desired weight region for any long period of time. This is what happens when your thermostat’s “optimal” level sits just to the right of the actual optimal calories range and closer to weight gain, leading to small but constant weight increase.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*A25WIJei2Fze6o2y0Rxmjg.jpeg" /><figcaption>When the “just right” of your appetite thermostat in fact sits at the edge of weight gain, making you systematically eat more than required (adapted from Walker, J. “The Hacker’s Diet”)</figcaption></figure><p>The restrictive phases weren’t dramatic crash diets, just periods of slightly heightened food awareness. The release phases weren’t obvious binges, but gradual returns to previous patterns. The oscillations were subtle enough that I’d attributed them to external factors rather than recognising the underlying pattern.</p><p>This recognition was humbling and liberating. Humbling because it revealed how little I’d understood my own behaviour. Liberating because it reframed the problem from personal failure to the “appetite thermostat” malfunction that can be tackled.</p><h4>Reframing the Problem</h4><p>The most valuable insight wasn’t technical but psychological: understanding that my weight struggles represented a broken feedback system rather than personal inadequacy. This reframing removed enormous emotional baggage from the process.</p><p>Daily weight swings were no longer a judgment on my willpower. They became just inevitable noise that needed better interpretation — and I knew that as a data person I can do that. A lot of accompanying negative emotions have melted away once I stopped thinking about maintaining my weight as a test of character, and started looking at it as a problem to solve.</p><p>With this understanding, the solution became obvious: <strong>build an external, correctly-working hunger thermostat to replace my internal broken one.</strong></p><h4>Making Peace with Hunger</h4><p>There is however one truth you can’t run away from. To lose weight I would still need to be in a calorie deficit, which would mean accepting that some hunger is inevitable and necessary during weight reduction. My thermostat would be asking for more calories, whereas I would deliberately keep it in a hungrier zone to lose weight.</p><p>But here’s what helped: remembering that this temporary discomfort was teaching me something valuable about why I shouldn’t overeat in the first place. Going through this necessary hunger journey would be motivation enough to avoid having to repeat it.</p><p>The real goal wasn’t to diet forever but to reset my normal. I would essentially retrain myself to be satisfied with less food — the amount that would maintain my target weight rather than my current one. The hunger would fade as my appetite adjusted to this new, better reality.</p><p>This changed everything about how I approached it. Instead of seeing it as a temporary punishment, it became a permanent upgrade to how I wanted to live — which made the whole process feel worthwhile rather than just something to endure. And hopefully this would lead to a way more sustainable approach than dieting.</p><h3>The Tools</h3><p>The suggested solution to a good external thermostat has two key parts: a way to see through daily weight noise, and clear feedback about whether you’re on track.</p><h4>Seeing Through the Noise</h4><p>If you recall the “water balloons with opinions” insight from earlier, this part of the solution helps filter out the daily weight fluctuations noise and isolate the actual underlying weight information that is useful.</p><p>To cut through the chaos, we need to smooth out those wild swings in the data. It’s a very similar method to the ones used in finance if you want to ignore the daily ups-and-downs of stock prices and reveal a slower moving real trend.</p><figure><img alt="The swarm of daily noise becomes a clear trend. You don’t want to have too many daily weigh-ins above the trend line so it doesn’t reverse." src="https://cdn-images-1.medium.com/max/1024/1*FKRy5S5lCCqI_58aJk_hfQ.png" /><figcaption>Daily weigh-in noise becomes clear pattern after making the trend line. The key is to watch out for too many daily observations above the trend line — that can start pushing it — and your weight — back up.</figcaption></figure><p>Once you have that smoothed trend line, everything becomes much more clear. Individual daily measurements can hover above or below the trend line, but what matters is how the trend line is evolving. If there are a lot of dots above the line, eventually they will start to “nudge the trend up”. If all your daily points are below the trend, it’s as if they are “pulling it down”. For weight loss, you want more days pulling the trend down than pushing it up. The occasional spikes above the line are absolutely fine and can be due to all kinds of reasons unrelated to the weight loss efforts. It’s the overall longer term smoothed pattern that tells the real story.</p><h4>Using this thermostat to adjust your behaviour</h4><p>This feedback system is beautifully simple: if there are too many daily measurements in a row that sit above the trend line, eat a bit less going forward. Below the line means you’re on track. No complex rules, no moral judgment, just clear visual feedback about whether your current approach is working.</p><p>I found myself thinking of each daily measurement as either helping or hindering my progress. Most days were neutral, but when I saw several dots above the trend line in a row, I knew it was time to pay more attention to my eating again.</p><h4>Getting Started</h4><p>Walker’s <a href="https://www.fourmilab.ch/hackdiet/">original book</a> includes spreadsheet templates that do all the trend calculations for you — perfectly adequate and simple to use for anyone willing to put some effort in.</p><p>But if you’re the type who enjoys tinkering with a bit of code, I’ve put together <a href="https://medium.com/@dmescherina/how-i-lost-weight-for-good-with-data-science-part-2-the-tools-86eb2ab5c558">a more detailed technical companion piece</a> with Python code to build your own interactive dashboard.</p><h3>Results</h3><p>In short: it worked for me!</p><figure><img alt="I made this programmatic thermostat early March, and the system started to pay off from late March onwards!" src="https://cdn-images-1.medium.com/max/1024/1*q2mgOjiUeqIrS9qzE4Zt7w.png" /><figcaption>You can see the trend has shifted from late March after I started to use my “external hunger thermostat” consistently.</figcaption></figure><p>For months before March, I was trapped in the Oscillating Oscar pattern, gaining and losing the same weight repeatedly. I’d have a few good weeks, lose a bit, then gradually drift back up. Because these swings were always subtle enough, I never really saw the actual trend until I started plotting the data and seeing the smoothed trend line.</p><p>Once I began using the external thermostat in earnest, something shifted. Not immediately — this isn’t a dramatic transformation story — but the oscillation pattern that had defined my relationship with weight finally broke. Instead of that frustrating cycle of temporary progress followed by inevitable regression, the trend line started pointing steadily downward.</p><h4>Small Corrections, Steady Results</h4><p>You can see it was not all smooth journey, and there were periods when I got off track. However, the beauty of having a clear trend line was being able to make tiny adjustments before problems escalated. In my old oscillating pattern, I’d ignore gradual weight rise, then panic and overcorrect with dramatic restrictions. Now, when I saw three or four dots above the red line in a row, I’d simply pay more attention to portion sizes for a few days. No all-or-nothing thinking, just gentle course corrections.</p><p>This matched perfectly with Walker’s feedback system concept. My broken internal thermostat had not provided the correct timely signals, but the external visual feedback gave me exactly what I needed: clear, objective information about whether my current approach was working.</p><h4>Fat mass vs. lean mass</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ic4iJ6GFHLSl8zJjFNx3GA.png" /><figcaption>Similar dashboard for body fat helped to confirm the source of weight loss</figcaption></figure><p>Having body fat measurements from my scale turned out to be very helpful in this approach too. When you’re dieting, there’s always the worry that you’re losing muscle along with fat during caloric restriction.</p><p>The fat percentage data became my reassurance that the weight loss was coming from the right places. During several plateau periods where my total weight barely budged for weeks, I could see that my body fat percentage was still dropping steadily. This meant the “stalled” weight loss was actually continued fat loss, just masked by water fluctuations or minor muscle retention. This transformed what could have been discouraging stagnation into evidence of success. The scale might lie by omission about day-to-day progress, but the combination of weight trend plus fat percentage told the real story about what was happening.</p><h4>Sustaining the Change</h4><p>So far so good — but the real test would be if I can sustain this progress. I’m cautiously optimistic, as this approach seems different to anything I tried before.</p><p>The key difference is sustainability. I’m not suffering through extreme restrictions or introducing anything dramatic. I’m eating slightly less than my broken hunger thermostat wants, accepting some very mild hunger, and trusting the external feedback system to guide adjustments. It’s gradual by design — I deliberately chose a modest caloric deficit to avoid the unsustainable cycle that had trapped me before. This process now is also a learning experience, teaching me how to eat to maintain my desired weight once I get there. Who knows — maybe I won’t even need the external hunger thermostat anymore once all this becomes a habit.</p><h4>The End?</h4><p>If you’re someone who appreciates bringing analytical thinking to personal challenges, this framework might resonate with you and could become a useful tool. If not, I hope this was at least an entertaining glimpse into one person’s attempt to debug their relationship with food using spreadsheets and trend lines.</p><p>For me, the biggest change was learning to work with my limitations rather than against them. My internal thermostat may be permanently broken, but that doesn’t mean I can’t maintain a healthy weight — I just need better tools than intuition and willpower.</p><p>Here’s to finding new uses for old tools!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e70470fc2a2b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/in-fitness-and-in-health/how-i-lost-weight-for-good-with-data-science-part-1-the-mindset-e70470fc2a2b">How I Lost Weight for Good with Data Science: Part 1, The Mindset</a> was originally published in <a href="https://medium.com/in-fitness-and-in-health">Second Nature (formerly In Fitness &amp; In Health)</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Lost Weight for Good with Data Science: Part 2, The Tools]]></title>
            <link>https://medium.com/in-fitness-and-in-health/how-i-lost-weight-for-good-with-data-science-part-2-the-tools-86eb2ab5c558?source=rss-6471f9151c6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/86eb2ab5c558</guid>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[wellness]]></category>
            <category><![CDATA[weight-loss]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[fat-loss]]></category>
            <dc:creator><![CDATA[Daria Meshcherina]]></dc:creator>
            <pubDate>Sun, 18 Jan 2026 22:11:09 GMT</pubDate>
            <atom:updated>2026-01-22T17:25:57.347Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZsAD1OOn622kBlwrXkBWhw.png" /></figure><h4>Why I stopped trusting willpower and started using trend lines</h4><p>In <a href="https://medium.com/@dmescherina/how-i-lost-weight-with-data-science-part-1-the-mindset-e70470fc2a2b">Part 1</a>, we explored the conceptual framework: understanding that daily weight fluctuations are mostly noise, recognising broken appetite thermostat, and accepting that we need an external feedback system to guide our eating decisions. The theory was liberating, but theory alone doesn’t move trend lines.</p><p>Before diving into the implementation, it’s worth reinforcing why this approach works. Humans are essentially sophisticated water containers — we’re roughly 60% water, and that water weight swings wildly based on sodium intake, sleep quality, stress levels, and general entropy in our lives. This means daily weight measurements contain far more noise than signal.</p><p>Meanwhile, our appetite regulation system — the internal “thermostat” that should automatically adjust our food intake to maintain steady weight — is running Stone Age software in a modern environment. For most of human history, this system worked adequately because food scarcity provided natural constraints. Our regulatory mechanisms evolved to help us survive famines, not navigate constant caloric abundance. The result is predictably chaotic: a system that kept our ancestors alive now struggles with 24/7 food availability and hyperpalatable processed foods.</p><p>The solution to this conundrum is to create a better external hunger thermostat to replace our inner one that’s not working. The “external thermostat” doesn’t have to be some abstract concept — it’s a specific set of tools that filter signal from noise and provide clear visual feedback about whether your current approach to weight loss is working. If you’re someone who enjoys building your own solutions, this implementation gives you complete control over your data and analysis.</p><p>The best part? You don’t need to be a data scientist to use this. Whether you’re comfortable with Python or just getting started, I’ll walk you through everything you need to create your own weight tracking dashboard.</p><h3>Getting Started: Choose Your Path</h3><p>Below are a few options for complete flexibility. You can start simple and add complexity as needed, or dive straight into the full implementation if you’re technically inclined.</p><h4>Path 1: Complete Beginner (No coding experience)</h4><p>If Python feels intimidating, start with the spreadsheet version mentioned in <a href="https://www.fourmilab.ch/hackdiet/">Walker’s original book</a>. It does similar exponential smoothing calculations and gives you the visual feedback you need. You can always upgrade to Python version later when you’re curious about customisation.</p><h4>Path 2: Curious Beginner (Some familiarity with Python)</h4><p>You’ll need:</p><ul><li>Python installed on your computer</li><li>Jupyter Notebook (or any Python environment)</li><li>About 15 minutes to set everything up</li></ul><h4>Path 3: Data-Savvy User (Comfortable with Python)</h4><p>You can use this as a foundation and extend it however you like — add new metrics, connect to APIs, build web dashboards, or integrate with other health tracking tools.</p><h3>Quick Setup Guide</h3><h4>Installing Requirements</h4><p>If you’re new to Python, the easiest way is to install Anaconda, which includes everything you need. Otherwise, you’ll need these packages:</p><pre>pip install pandas plotly ipywidgets</pre><p>For Jupyter notebook users, you might also need:</p><pre>jupyter nbextension enable --py widgetsnbextension</pre><h3>Preparing Your Data</h3><p>Create a CSV file with your daily measurements. The format is deliberately simple:</p><pre>date,weight,fat<br>2024-01-01,70.2,22.5<br>2024-01-02,70.8,22.3<br>2024-01-03,69.9,22.7</pre><p><strong>Tips for data collection:</strong></p><ul><li>Weigh yourself at the same time each day (preferably morning, after using the bathroom)</li><li>Use any date format that makes sense to you — the code handles most common formats automatically</li><li>The fat percentage is optional but valuable if your scale provides it</li><li>Don’t worry about missing days — the code handles gaps</li><li>Weight units are auto-detected (kg if max weight &lt; 200, otherwise pounds)</li></ul><h3>The Technical Implementation</h3><h4>Core Concept: Exponentially Weighted Moving Average</h4><p>The magic happens through signal processing. Instead of getting distracted by daily fluctuations, we use an Exponentially Weighted Moving Average (EWMA) to reveal the underlying trend. This technique gives more importance to recent measurements while smoothing out the noise.</p><p>The formula is elegantly simple:</p><pre>Today&#39;s trend = α × Today&#39;s weight + (1-α) × Yesterday&#39;s trend</pre><p>The α (alpha) parameter controls sensitivity. I use α = 0.15, which means today’s measurement contributes 15% to the trend, while the previous trend contributes 85%. This creates a line that responds to real changes but ignores daily volatility.</p><h3>The WeightTracker Class</h3><p>Here’s the complete implementation. Don’t let the length intimidate you — most of it handles data formatting and creates interactive charts:</p><pre>import pandas as pd<br>import plotly.graph_objects as go<br>from datetime import datetime, timedelta<br><br>class WeightTracker:<br>    def __init__(self, csv_file):<br>        &quot;&quot;&quot;<br>        Initialize with CSV containing: date, weight, fat (optional)<br>        Date format: YYYY-MM-DD or DD/MM/YYYY<br>        Weight in kg or lbs (auto-detected)<br>        &quot;&quot;&quot;<br>        self.df = pd.read_csv(csv_file)<br>        self._preprocess_data()<br>    <br>    def _preprocess_data(self):<br>        # Convert date column to datetime<br>        self.df[&#39;date&#39;] = pd.to_datetime(self.df[&#39;date&#39;])<br>        self.df = self.df.sort_values(&#39;date&#39;)<br>        <br>        # Auto-detect weight unit (assume kg if max &lt; 200, else lbs)<br>        if self.df[&#39;weight&#39;].max() &gt; 200:<br>            self.df[&#39;weight_kg&#39;] = self.df[&#39;weight&#39;] / 2.20462<br>            self.df[&#39;weight_lb&#39;] = self.df[&#39;weight&#39;]<br>        else:<br>            self.df[&#39;weight_kg&#39;] = self.df[&#39;weight&#39;]<br>            self.df[&#39;weight_lb&#39;] = self.df[&#39;weight&#39;] * 2.20462<br>    <br>    def filter_by_date(self, start_date=None, end_date=None, days_back=None):<br>        &quot;&quot;&quot;Filter dataframe by date range&quot;&quot;&quot;<br>        df_filtered = self.df.copy()<br>        <br>        if days_back is not None:<br>            latest_date = df_filtered[&#39;date&#39;].max()<br>            cutoff_date = latest_date - pd.Timedelta(days=days_back)<br>            df_filtered = df_filtered[df_filtered[&#39;date&#39;] &gt;= cutoff_date]<br>        <br>        if start_date:<br>            if hasattr(start_date, &#39;strftime&#39;):<br>                start_date = pd.Timestamp(start_date)<br>            df_filtered = df_filtered[df_filtered[&#39;date&#39;] &gt;= start_date]<br>        <br>        if end_date:<br>            if hasattr(end_date, &#39;strftime&#39;):<br>                end_date = pd.Timestamp(end_date)<br>            df_filtered = df_filtered[df_filtered[&#39;date&#39;] &lt;= end_date]<br>        <br>        return df_filtered<br>    <br>    def plot_weight_trend(self, alpha=0.15, start_date=None, end_date=None, <br>                         days_back=None, unit=&#39;kg&#39;):<br>        &quot;&quot;&quot;Create interactive weight trend visualisation&quot;&quot;&quot;<br>        df = self.filter_by_date(start_date, end_date, days_back)<br>        <br>        if len(df) == 0:<br>            print(&quot;No data in selected date range&quot;)<br>            return None<br>        <br>        weight_col = &#39;weight_lb&#39; if unit == &#39;lb&#39; else &#39;weight_kg&#39;<br>        trend = df[weight_col].ewm(alpha=alpha, adjust=False).mean()<br>        <br>        fig = go.Figure()<br>        <br>        # Raw weight data<br>        fig.add_trace(go.Scatter(<br>            x=df[&#39;date&#39;], y=df[weight_col],<br>            mode=&#39;markers+lines&#39;,<br>            line=dict(color=&#39;rgba(0,0,255,0.3)&#39;, width=1),<br>            marker=dict(size=6, color=&#39;blue&#39;),<br>            name=f&#39;Weight ({unit})&#39;<br>        ))<br>        <br>        # Trend line<br>        fig.add_trace(go.Scatter(<br>            x=df[&#39;date&#39;], y=trend,<br>            mode=&#39;lines&#39;,<br>            line=dict(color=&#39;red&#39;, width=2),<br>            name=f&#39;Trend (α={alpha:.2f})&#39;<br>        ))<br>        <br>        # Calculate stats<br>        total_change = df[weight_col].iloc[-1] - df[weight_col].iloc[0]<br>        days = (df[&#39;date&#39;].iloc[-1] - df[&#39;date&#39;].iloc[0]).days or 1<br>        monthly_rate = (total_change / days) * 30<br>        <br>        # Add stats annotation<br>        fig.add_annotation(<br>            x=0.02, y=0.98, xref=&quot;paper&quot;, yref=&quot;paper&quot;,<br>            text=f&quot;Total: {total_change:.1f} {unit}&lt;br&gt;&quot;<br>                 f&quot;Monthly rate: {monthly_rate:.2f} {unit}&quot;,<br>            showarrow=False, bgcolor=&quot;white&quot;, <br>            bordercolor=&quot;black&quot;, borderwidth=1<br>        )<br>        <br>        # Layout with range selector<br>        fig.update_layout(<br>            title=f&#39;Weight Trend with EWMA (α={alpha:.2f})&#39;,<br>            xaxis_title=&#39;Date&#39;, yaxis_title=f&#39;Weight ({unit})&#39;,<br>            hovermode=&quot;x unified&quot;, height=600,<br>            xaxis=dict(<br>                rangeselector=dict(<br>                    buttons=list([<br>                        dict(count=1, label=&quot;1m&quot;, step=&quot;month&quot;, stepmode=&quot;backward&quot;),<br>                        dict(count=3, label=&quot;3m&quot;, step=&quot;month&quot;, stepmode=&quot;backward&quot;),<br>                        dict(count=6, label=&quot;6m&quot;, step=&quot;month&quot;, stepmode=&quot;backward&quot;),<br>                        dict(count=1, label=&quot;1y&quot;, step=&quot;year&quot;, stepmode=&quot;backward&quot;),<br>                        dict(step=&quot;all&quot;)<br>                    ])<br>                ),<br>                rangeslider=dict(visible=True),<br>                type=&quot;date&quot;<br>            )<br>        )<br>        <br>        return fig<br>    <br>    def plot_fat_trend(self, alpha=0.15, start_date=None, end_date=None, days_back=None):<br>        &quot;&quot;&quot;Create body fat trend if fat data available&quot;&quot;&quot;<br>        if &#39;fat&#39; not in self.df.columns:<br>            print(&quot;No fat percentage data available&quot;)<br>            return None<br>        <br>        df = self.filter_by_date(start_date, end_date, days_back)<br>        df = df.dropna(subset=[&#39;fat&#39;])<br>        <br>        if len(df) == 0:<br>            print(&quot;No fat data in selected range&quot;)<br>            return None<br>        <br>        # Calculate fat mass in kg<br>        df[&#39;fat_kg&#39;] = df[&#39;weight_kg&#39;] * df[&#39;fat&#39;] / 100<br>        fat_trend = df[&#39;fat_kg&#39;].ewm(alpha=alpha, adjust=False).mean()<br>        <br>        fig = go.Figure()<br>        <br>        fig.add_trace(go.Scatter(<br>            x=df[&#39;date&#39;], y=df[&#39;fat_kg&#39;],<br>            mode=&#39;markers+lines&#39;,<br>            line=dict(color=&#39;rgba(0,128,0,0.3)&#39;, width=1),<br>            marker=dict(size=6, color=&#39;green&#39;),<br>            name=&#39;Body Fat (kg)&#39;<br>        ))<br>        <br>        fig.add_trace(go.Scatter(<br>            x=df[&#39;date&#39;], y=fat_trend,<br>            mode=&#39;lines&#39;,<br>            line=dict(color=&#39;darkgreen&#39;, width=2),<br>            name=f&#39;Fat Trend (α={alpha:.2f})&#39;<br>        ))<br>        <br>        # Calculate fat loss stats<br>        total_fat_change = df[&#39;fat_kg&#39;].iloc[-1] - df[&#39;fat_kg&#39;].iloc[0]<br>        days = (df[&#39;date&#39;].iloc[-1] - df[&#39;date&#39;].iloc[0]).days or 1<br>        monthly_fat_rate = (total_fat_change / days) * 30<br>        <br>        fig.add_annotation(<br>            x=0.02, y=0.98, xref=&quot;paper&quot;, yref=&quot;paper&quot;,<br>            text=f&quot;Fat change: {total_fat_change:.1f} kg&lt;br&gt;&quot;<br>                 f&quot;Monthly rate: {monthly_fat_rate:.2f} kg&quot;,<br>            showarrow=False, bgcolor=&quot;white&quot;,<br>            bordercolor=&quot;black&quot;, borderwidth=1<br>        )<br>        <br>        fig.update_layout(<br>            title=&#39;Body Fat Trend (kg)&#39;,<br>            xaxis_title=&#39;Date&#39;, yaxis_title=&#39;Body Fat (kg)&#39;,<br>            hovermode=&quot;x unified&quot;, height=600,<br>            xaxis=dict(<br>                rangeselector=dict(<br>                    buttons=list([<br>                        dict(count=1, label=&quot;1m&quot;, step=&quot;month&quot;, stepmode=&quot;backward&quot;),<br>                        dict(count=3, label=&quot;3m&quot;, step=&quot;month&quot;, stepmode=&quot;backward&quot;),<br>                        dict(count=6, label=&quot;6m&quot;, step=&quot;month&quot;, stepmode=&quot;backward&quot;),<br>                        dict(step=&quot;all&quot;)<br>                    ])<br>                ),<br>                rangeslider=dict(visible=True),<br>                type=&quot;date&quot;<br>            )<br>        )<br>        <br>        return fig<br><br>def create_dashboard(csv_file):<br>    &quot;&quot;&quot;Create interactive dashboard with weight and fat trends&quot;&quot;&quot;<br>    try:<br>        import ipywidgets as widgets<br>        from IPython.display import display, clear_output<br>        from plotly.offline import iplot<br>        <br>        tracker = WeightTracker(csv_file)<br>        <br>        # Create controls<br>        unit_dropdown = widgets.Dropdown(<br>            options=[&#39;kg&#39;, &#39;lb&#39;], value=&#39;kg&#39;, description=&#39;Unit:&#39;,<br>            layout=widgets.Layout(width=&#39;150px&#39;)<br>        )<br>        <br>        # Create tab outputs<br>        weight_output = widgets.Output(layout=widgets.Layout(height=&#39;650px&#39;))<br>        fat_output = widgets.Output(layout=widgets.Layout(height=&#39;650px&#39;))<br>        <br>        # Create tabs<br>        tab = widgets.Tab()<br>        tab.children = [weight_output, fat_output]<br>        tab.set_title(0, &#39;Weight Trend&#39;)<br>        tab.set_title(1, &#39;Fat Trend&#39;)<br>        <br>        def update_plots(unit):<br>            # Update weight tab<br>            with weight_output:<br>                clear_output(wait=True)<br>                try:<br>                    weight_fig = tracker.plot_weight_trend(alpha=0.15, unit=unit)<br>                    if weight_fig:<br>                        iplot(weight_fig)<br>                    else:<br>                        print(&quot;Error creating weight plot&quot;)<br>                except Exception as e:<br>                    print(f&quot;Weight plot error: {str(e)}&quot;)<br>            <br>            # Update fat tab<br>            with fat_output:<br>                clear_output(wait=True)<br>                try:<br>                    if &#39;fat&#39; not in tracker.df.columns or tracker.df[&#39;fat&#39;].isna().all():<br>                        print(&quot;No fat percentage data available&quot;)<br>                        print(&quot;\nTo add fat tracking, include it in your CSV:&quot;)<br>                        print(&quot;date,weight,fat&quot;)<br>                        print(&quot;2024-01-01,70.2,22.5&quot;)<br>                    else:<br>                        fat_fig = tracker.plot_fat_trend(alpha=0.15)<br>                        if fat_fig:<br>                            iplot(fat_fig)<br>                        else:<br>                            print(&quot;Error creating fat plot&quot;)<br>                except Exception as e:<br>                    print(f&quot;Fat plot error: {str(e)}&quot;)<br>        <br>        # Create interactive interface<br>        interactive_plot = widgets.interactive(update_plots, unit=unit_dropdown)<br>        dashboard = widgets.VBox([unit_dropdown, tab])<br>        display(dashboard)<br>        <br>        # Initial update<br>        update_plots(&#39;kg&#39;)<br>        <br>    except ImportError:<br>        print(&quot;Dashboard requires: pip install ipywidgets plotly&quot;)<br>        print(&quot;Running basic version instead...&quot;)<br>        tracker = WeightTracker(csv_file)<br>        fig = tracker.plot_weight_trend()<br>        fig.show()</pre><h3>Usage Examples</h3><h4>Basic Usage</h4><pre># Load your data and create basic plots<br>tracker = WeightTracker(&#39;my_weight_data.csv&#39;)<br><br># Simple weight trend<br>fig = tracker.plot_weight_trend()<br>fig.show()<br><br># Look at just the last 3 months<br>fig = tracker.plot_weight_trend(days_back=90)<br>fig.show()<br><br># Fat trend (if you have fat percentage data)<br>fat_fig = tracker.plot_fat_trend()<br>fat_fig.show()</pre><h4>Interactive Dashboard</h4><pre># Create full interactive dashboard (works in Jupyter)<br>create_dashboard(&#39;my_weight_data.csv&#39;)</pre><h4>Customisations</h4><p><strong>Adjusting sensitivity:</strong> If the trend line feels too sluggish or too reactive, experiment with the alpha parameter:</p><pre># More responsive to recent changes<br>fig = tracker.plot_weight_trend(alpha=0.25)<br><br># More stable, less influenced by recent fluctuations<br>fig = tracker.plot_weight_trend(alpha=0.10)</pre><p><strong>Focusing on specific periods:</strong> When you want to analyse particular time ranges:</p><pre># Look at progress since starting your new approach<br>fig = tracker.plot_weight_trend(start_date=&#39;2024-03-01&#39;)<br><br># Compare different months<br>march_fig = tracker.plot_weight_trend(start_date=&#39;2024-03-01&#39;, end_date=&#39;2024-03-31&#39;)</pre><h4>Common Issues and Solutions</h4><p><strong>“No module named ‘plotly’”</strong>: Install the required packages with pip install pandas plotly ipywidgets</p><p><strong>Dates not displaying correctly</strong>: The code handles most date formats automatically, but if you’re having issues, stick to YYYY-MM-DD format in your CSV.</p><p><strong>Interactive widgets not working</strong>: In Jupyter, run jupyter nbextension enable --py widgetsnbextension and restart your kernel.</p><p><strong>Charts not displaying</strong>: If you’re not using Jupyter, call .show() on the figure object to display charts in your browser.</p><h4>Making It Your Own</h4><p>This implementation is designed to be a foundation rather than a finished product. Some extensions I’ve considered but so far have found excessive:</p><ul><li><strong>Integration with smart scales</strong>: Connect directly to your scale’s API to automate data collection</li><li><strong>Additional metrics</strong>: Track sleep, exercise, or other factors that might correlate with weight changes</li><li><strong>Goal tracking</strong>: Add target weight lines and progress indicators</li><li><strong>Alerts</strong>: Get notifications when you’ve been above/below trend for several days</li></ul><p>The code is structured to make these additions straightforward — you’re working with standard pandas DataFrames and plotly figures, so integration with other tools should be simple.</p><h3>The Real Test: Daily Use</h3><p>The technical implementation is only valuable if it actually changes your behaviour. The visual feedback system needs to become part of your routine, not just an interesting exercise in data analysis (although it was for me!)</p><p>I’ve found that checking the dashboard weekly gives me the information I need without becoming obsessive about daily fluctuations. When I see several dots above the trend line, I know it’s time to pay attention to portion sizes for a few days. When everything’s tracking downward, I can relax and trust the process.</p><p>This approach isn’t about perfection — it’s about having better information to make small, sustainable adjustments. The external thermostat doesn’t tell you what to eat or when to exercise. It simply shows you whether your current approach is working, so you can make informed decisions about what to change.</p><p>Here’s to finding new practical applications for the old tools!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=86eb2ab5c558" width="1" height="1" alt=""><hr><p><a href="https://medium.com/in-fitness-and-in-health/how-i-lost-weight-for-good-with-data-science-part-2-the-tools-86eb2ab5c558">How I Lost Weight for Good with Data Science: Part 2, The Tools</a> was originally published in <a href="https://medium.com/in-fitness-and-in-health">Second Nature (formerly In Fitness &amp; In Health)</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>