<?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 Saachi Subudhi on Medium]]></title>
        <description><![CDATA[Stories by Saachi Subudhi on Medium]]></description>
        <link>https://medium.com/@saachi.subudhi2011?source=rss-5f80c6a269b7------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*uytAeUSammrQ7-3B0aYD2A.png</url>
            <title>Stories by Saachi Subudhi on Medium</title>
            <link>https://medium.com/@saachi.subudhi2011?source=rss-5f80c6a269b7------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 17:43:20 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@saachi.subudhi2011/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[The Organ Textbooks Forgot: What the Rete Ovarii Reveals About the Gender Data Gap]]></title>
            <link>https://medium.com/@saachi.subudhi2011/the-organ-textbooks-forgot-what-the-rete-ovarii-reveals-about-the-gender-data-gap-b5faf08802ec?source=rss-5f80c6a269b7------2</link>
            <guid isPermaLink="false">https://medium.com/p/b5faf08802ec</guid>
            <category><![CDATA[gender]]></category>
            <category><![CDATA[rete-ovarii]]></category>
            <category><![CDATA[gender-bias]]></category>
            <dc:creator><![CDATA[Saachi Subudhi]]></dc:creator>
            <pubDate>Mon, 26 Oct 2026 14:14:32 GMT</pubDate>
            <atom:updated>2026-10-29T17:53:10.806Z</atom:updated>
            <content:encoded><![CDATA[<h3>“What happens when medicine mistakes what it does not understand for something that does not matter?” my friend asked during recess.</h3><h3>“Neglect.” I answered.</h3><p>Let’s time travel back to the 1870s when anatomists described a small network of tubules attached to the ovary. This organ is called the rete ovarii. It appeared in drawings in early editions of Gray’s Anatomy, then gradually vanished from modern textbooks, implied as a functionless organ.</p><p><em>Functionless? More like dismissed.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iRWcVGGgYrkip1z5mpf4ww.png" /><figcaption>Image of labelled ovary from Wikimedia Commons</figcaption></figure><p>In recent years, developmental biologists have brought it back. Imaging work in mice, including a 2022 study that visualized it in whole ovaries, drew new attention to the structure. A follow-up study in <a href="https://elifesciences.org/articles/96662"><em>eLife</em> by <em>Dilara Anbarci</em></a><em> </em>and colleagues suggested it is not inert: it appears to channel protein-rich fluid into the ovary and may respond to hormonal signals, leading some researchers to compare it to an antenna. This is crucial. However, this is mouse research, and its role in adult ovarian function is still unclear, and some reviewers felt parts of the evidence were not yet conclusive. Regardless of its human significance, this raises an uncomfortable question: if part of the ovary could be overlooked for more than a century, what else has been missed?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1021/1*-QY09zVR4XyXc9mlJRotRQ.png" /><figcaption>The rete ovarri extends from the ovary and curves around the organ like a horse shoe. Credit: Dilara Anbarci/<em>eLife </em>2025</figcaption></figure><h3><strong>A chain, not a single gap.</strong></h3><p>The rete ovarii is just one example of a large pattern known as the gender data gap: the systematic shortage of research, data, and analysis that are related to women’s bodies. The gap is not one failure but a chain of failures.</p><p><em>Which is heartbreaking to think of in concerns for the future generations.</em> <em>Are they at risk? Are we at risk?</em></p><p>What isn’t studied in the lab shapes what is tested in trials, what is taught to medical students, who are our future doctors, and ultimately how patients are treated. Each link inherits the blind spots of the one before it. Seen this way, the question is less about a single organ and more about how knowledge is built, and whose bodies it is built on. This raises the question of the fate of future usage of artificial intelligence in healthcare, as it is built on these very patterns, leading to long-term consequences.</p><h3><strong>The first link: the laboratory.</strong></h3><p>The chain starts in preclinical research. For decades, many studies used male animals and male cells as the default, partly for convenience and partly because researchers worried that female hormonal cycles would add variability to results. However, male animals also show hormonal fluctuations, and studies found their results aren’t necessarily less variable.</p><p><em>A bit ridiculous, don’t you think?</em></p><p>As a result, a great deal of foundational knowledge about disease and drug response comes from male-only experiments. In response, in 2016, <a href="https://orwh.od.nih.gov/sex-as-biological-variable">the US National Institutes of Health</a> began expecting researchers to consider sex as a biological variable in their study designs. Even so, policy change takes time to work through the research pipeline, and earlier assumptions are still embedded in the literature and in the models scientists use today. Cases like the rete ovarii showcase how easily a structure can be labelled unimportant and then go unexamined, simply because nobody looked close enough.</p><h3><strong>The second link: clinical trials.</strong></h3><p>Human trials. In the decades after the thalidomide tragedy, regulators and companies grew cautious about including women of childbearing potential, leaving women underrepresented in clinical trials.</p><blockquote>Imagine a pregnant woman prescribed a medicine that was never tested on pregnant women, so neither she nor her doctor knows how it will affect her or her baby.</blockquote><p>In the United States, the <a href="https://www.congress.gov/bill/103rd-congress/senate-bill/1">1993 NIH Revitalization Act</a> required the inclusion of women and minorities in NIH-funded clinical research, a major step forward. Yet formal inclusion is not the same as equal evidence. Participation can still lag in some fields. Pregnant and lactating women are routinely excluded, leaving real gaps in medical knowledge about medicines they need. And even when women are enrolled, results often hide differences between men and women in how well a treatment works or what side effects it causes. The deeper challenge has shifted from inclusion to analysis: counting women is not enough if the data are never broken down by sex.</p><p>The consequences are real. Because many drugs were developed and dosed using mostly male data, women have experienced higher rates of adverse reactions to some medicines. <a href="https://www.fda.gov/media/84992/download">A frequently cited example is the insomnia drug zolpidem, for which US regulators in January 2013 </a>recommended a lower starting dose for women after evidence showed higher drug levels in their bodies the next morning. The pharmacokinetic trials of 10 mg Ambien lead to 15% of 250 women having zolpidem concentrations that exceeded 50 ng/ml 8 hours post intake, which may lead to slower reactions and impaired coordination when driving, leading to increased risk of road accidents. The dosage was reduced from 10 mg to 5 mg for products like Zolpimist, Edluar, and Ambien, and for Ambien CR the dosage reduced from 12.5 mg — 6.25 mg.</p><h3><strong>The third link: how pain is judged.</strong></h3><p>Unfortunately, this chain does not end in the laboratory or the trial. It continues into the clinic, where research gaps meet human judgment. One of the best-known studies is<a href="https://digitalcommons.law.umaryland.edu/cgi/viewcontent.cgi?params=/context/fac_pubs/article/1144/&amp;path_info=The_Girl_who_Cried_Pain.pdf"> Diane Hoffmann and Anita Tarzian’s 2001 review, ‘The Girl Who Cried Pain.’</a> They found that women report more severe, more frequent, and longer-lasting pain than men, yet are often treated less aggressively. Among the explanations they discuss are cultural and I’d argue, somewhat misogynistic stereotypes about how pain should look and a tendency to attribute women’s pain to emotional or psychological causes, in some cases referred to as “hysteria.”<a href="https://pubmed.ncbi.nlm.nih.gov/29682130/"> A later review by Anke Samulowitz and colleagues in 2018</a>, covering 77 studies, again identified gender bias in pain treatment in both the patient-provider encounter and clinicians’ decisions.</p><figure><img alt="An extract from the Girl Who Cried Pain written by Diane E. Hoffmann and Anita J. Tarzian" src="https://cdn-images-1.medium.com/max/765/1*GE5JIj5zfg1kZ7Sw0Rh2lA.png" /><figcaption>An extract from the Girl Who Cried Pain written by Diane E. Hoffmann and Anita J. Tarzian</figcaption></figure><p><em>If so many studies have examined attitudes towards pain, why don’t women feel safer speaking up?</em></p><p>These findings should be read with care. Pain is subjective and difficult to measure; the usual out-of-10 scale is variable from person to person, and researchers stress that neither sex nor gender alone can explain observed differences: biology, social norms, and the clinical encounter all interact. Findings from laboratory pain experiments can also differ from what is seen in real-world care. But the broader point holds. Biased evidence feeds biased clinical judgement, and the two reinforce each other. When medical knowledge is built mostly around male bodies, women’s symptoms can look <em>“atypical”, </em>and atypical symptoms can slide into dismissal very easily.</p><h3><strong>Why the gap persists:</strong></h3><p>No single cause explains all this. Historical and cultural assumptions about women’s bodies and disease laid the groundwork of medicine and still influence research priorities. On top of that sit practical and structural factors: the convenience of single-sex (mostly male) study designs, caution around pregnancy, not enough funding for conditions that mainly affect women, the underrepresentation of women in some research leadership and decision-making roles, and slow change within universities and institutes. Dosing guidelines, textbooks, and protocols built on older data remain in use for students long after policies change. Barriers to participation, such as caregiving responsibilities and trial logistics, can also limit who actually enrolls. Cultural bias is best understood as a foundation on which these other factors build up rather than the main cause.</p><h3><strong>The next chapter: artificial intelligence.</strong></h3><p>These patterns matter more as medicine develops with artificial intelligence. Machine-learning systems learn from existing data: health records, imaging archives, trial results, and published literature. If those sources underrepresent women or leave out sex-specific research and analysis, the systems can inherit the same gaps and apply them at scale, which is horrifying to think of.</p><p>A related problem has already been shown in medical imaging. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC7293650">A 2020 study by Agostina Larrazabal and colleagues </a>found that chest X-ray classifiers trained on datasets imbalanced by sex performed significantly worse for the underrepresented sex. Their main public dataset had 112,120 chest X-rays from 30,805 patients, covering 14 diseases, and it is itself imbalanced: 56.5% male, 43.5% female images. Imagine a hospital in 2045 that uses an AI tool to screen chest X-rays for early signs of lung disease.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QR3gBCj73L9cq53_qdDTzg.png" /><figcaption>Some example of images from the public databases ( source 11)</figcaption></figure><p>The tool was trained mostly on scans from men, but its overall accuracy looks excellent, so nobody checks how it performs for women. A woman’s scan is read as “clear,” the early signs are missed, and she is sent home. Months later, her disease is found at a more advanced stage. No one blames the software, because on paper it is accurate. The lesson is not that AI is inevitably biased. If used well, it could help identify gaps, flag differences in outcomes, and bring overlooked research to light. However, it will not correct biased data on its own. For today’s children, who will grow up with algorithms as a normal part of healthcare, the quality and balance of the data being collected now will lay the foundation for the care they receive.</p><h3><strong>My Hot Take:</strong></h3><p><em>Medical AI tools should not be banned from hospitals even if they don’t publish accuracy separately for men and women.</em></p><p>I would say that instead of outrightly banning this technology, we should not let it work by itself. AI’s analysis should be rechecked and monitored by humans, and any diagnosis made by AI should go through thorough and multiple rounds of checking, especially when sex is a variable. Additionally, an AI tool that reports separate results for men and women is where we could start; however, a second reviewer, such as a doctor, is necessary. The checkers also need to be trained to look for sex differences, because we have already seen how human judgement has dismissed women’s symptoms. If medicine rediscovered the rete ovarii after more than a century, it can’t afford to hand its next blind spot to a machine and just stop looking.</p><p><strong>Some may argue:</strong></p><ol><li>AI tools have the ability to showcase and advertise an accuracy of 95% ; however, they might not work as well for a particular group of people. This was seen in the 2020 Larrazabal study even though women made up a quarter of the training data.</li><li>Hospitals can’t even see the results differentiated by sex so they wouldn’t be able to know if the AI tool is completely safe for their patients.</li><li>It took until 2013 for the FDA to realise the correct dose of zolpidem for women and halve the recommended dose, after years of dosing that nobody had analysed using sex as a variable.</li><li>Medical AI learns by thousands of labelled examples made by doctors, such as X-rays that are healthy or diseased, and it repeatedly adjusts its internal settings until its guesses match the labels. It only learns the patterns in the data given, and most of the data given is based on men, so AI’s diagnosis will fit men best. Therefore, its mistakes on women may go unnoticed unless someone tests for them separately and feeds separate data.</li></ol><p>Therefore, with AI, the same mistake could spread to thousands of patients before anyone notices.</p><p>However, a complete ban may also be harmful. Sex is only one of so many variables, such as age and ethnicity. Additionally, keeping medical AI tools out of hospitals could mean more missed diagnoses, which doesn’t benefit both doctor and patient, as it could still diagnose better than current practice for both men and women. Hospitals with fewer resources could struggle to provide sufficient data, which could slow innovation or push people towards unregulated tools. Furthermore, various public medical datasets don’t even record patient sex, so it would be expensive to make an entirely new data collection.</p><blockquote>The rete ovarii offers a final lesson: something dismissed as functionless may simply be unstudied. The question this article leaves you with is: “What can you do about it?”</blockquote><h3>About the Author:</h3><p>Hello! I am Saachi Subudhi. I am a student in Year 11 at GEMS Wellington International School, a member of TKS Dubai and an advocate for gender equality. I’m passionate about the history of gender inequality and its impacts across various categories. Learning about the very late rediscovery of the rete ovarii encouraged me to understand the cause and try to help stop this bias before it’s too late. Hope you learned something new from this article!</p><h3>Sources</h3><ul><li>Anbarci, D. et al. “Rediscovering the rete ovarii, a secreting auxiliary structure to the ovary.” <em>eLife</em>. <a href="https://elifesciences.org/articles/96662">https://elifesciences.org/articles/96662</a></li><li>Science News. “An overlooked organ may help the ovary function.” <a href="https://www.sciencenews.org/?p=3153125">https://www.sciencenews.org/?p=3153125</a></li><li>ZME Science. “Scientists rediscover a lost piece of female anatomy that may play a crucial role in fertility.” <a href="https://zmescience.com/science/news-science/scientists-rediscover-a-lost-piece-of-female-anatomy-that-may-play-a-crucial-role-in-fertility">https://zmescience.com/science/news-science/scientists-rediscover-a-lost-piece-of-female-anatomy-that-may-play-a-crucial-role-in-fertility</a></li><li>Hoffmann, D. E., &amp; Tarzian, A. J. (2001). “The Girl Who Cried Pain: A Bias Against Women in the Treatment of Pain.” <em>Journal of Law, Medicine &amp; Ethics</em>, 29(1), 13–27. doi:10.1111/j.1748–720x.2001.tb00037.x</li><li>Samulowitz, A., Gremyr, I., Eriksson, E., &amp; Hensing, G. (2018). “‘Brave Men’ and ‘Emotional Women’: A Theory-Guided Literature Review on Gender Bias in Health Care and Gendered Norms towards Patients with Chronic Pain.” <em>Pain Research and Management</em>, 2018:6358624. doi:10.1155/2018/6358624</li><li>Larrazabal, A. J., Nieto, N., Peterson, V., Milone, D. H., &amp; Ferrante, E. (2020). “Gender imbalance in medical imaging datasets produces biased classifiers for computer-aided diagnosis.” Proceedings of the National Academy of Sciences, 117(23), 12592–12594. doi:10.1073/pnas.1919012117. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC7293650/">https://pmc.ncbi.nlm.nih.gov/articles/PMC7293650</a></li><li>US National Institutes of Health. Policy on consideration of sex as a biological variable (2016), and the NIH Revitalization Act of 1993.</li><li>U.S. Food and Drug Administration (2013, January 10). “FDA Drug Safety Communication: Risk of next-morning impairment after use of insomnia drugs; FDA requires lower recommended doses for certain drugs containing zolpidem (Ambien, Ambien CR, Edluar, and Zolpimist).” <a href="https://www.fda.gov/media/84992/download">https://www.fda.gov/media/84992/download</a></li><li>Department of Health and Social Care (2022). “Results of the ‘Women’s Health — Let’s talk about it’ survey.” Women’s Health Strategy: Call for Evidence outcome, GOV.UK. First published 23 December 2021; updated 13 April 2022. <a href="https://www.gov.uk/government/calls-for-evidence/womens-health-strategy-call-for-evidence/outcome/results-of-the-womens-health-lets-talk-about-it-survey">https://www.gov.uk/government/calls-for-evidence/womens-health-strategy-call-for-evidence/outcome/results-of-the-womens-health-lets-talk-about-it-survey</a></li><li>U.S. Food and Drug Administration, Health Canada, &amp; Medicines and Healthcare products Regulatory Agency (2021). “Good machine learning practice for medical device development: guiding principles.” <a href="https://www.gov.uk/government/publications/good-machine-learning-practice-for-medical-device-development-guiding-principles/good-machine-learning-practice-for-medical-device-development-guiding-principles">https://www.gov.uk/government/publications/good-machine-learning-practice-for-medical-device-development-guiding-principles/good-machine-learning-practice-for-medical-device-development-guiding-principles</a></li><li>Larrazabal AJ, Nieto N, Peterson V, Milone DH, Ferrante E. Gender imbalance in medical imaging datasets produces biased classifiers for computer-aided diagnosis. Proc Natl Acad Sci U S A. 2020;117(23):12592–4. doi:10.1073/pnas.1919012117</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b5faf08802ec" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>