The Most Dangerous Myth in Medicine: The Average Patient
What an AI dosing error, a stock-market collapse, and a forgotten heart-drug study reveal about the future of precision health.
📌 Special note 1: This newsletter is free. But it takes hours each week to research, write, and produce at this level. Here are 5 ways to support my work: 1. click “❤️” to amplify 2. subscribe 3. share this publication 4. buy me coffee 5. become a partner.
📕Special note 2: The Billion Dollar Blindspot is almost ready to step into the world. If you want to follow the journey, early excerpts, behind-the-scenes notes, and launch news, you can sign up for early access here.
Connect and collaborate with me here! Email | LinkedIn
When Medical AI Gets the Rule Right and the Patient Wrong
Two days ago, infectious disease physician M. Rizwan Sohail shared a post on X that stopped me mid-scroll. He had been testing OpenEvidence, a clinical AI tool built to support physicians with medication decisions. One of the prompts was straightforward:
How should oxacillin be dosed in a dialysis patient?
The model responded with confidence: Reduce the dose.
The explanation looked authoritative. It referenced renal impairment, standard precautions, and the familiar logic that guides dosing for most drugs in this antibiotic class. Everything in the reply sounded correct, even textbook.
Except it wasn’t.
Oxacillin is the exception in its class. It does not need a dose reduction in dialysis patients1. Every infectious disease trainee learns this early, precisely because it breaks the pattern.
What makes this moment unsettling is that the AI wasn’t inventing a fantasy. There was no hallucination. It applied the right general rule to the one case where the rule does not apply. In AI, this is called semantic drift, a subtle shift in how a model interprets or applies knowledge, invisible until it lands on the wrong side of a decision.
The most dangerous AI errors aren’t inventions. They are misapplied truths.
In medicine, that kind of drift is not theoretical. It has a human cost.
New Here?
If you are an investor, operator, or women’s health advocate, start here.
→ Preorder the book: The Billion Dollar Blind Spot
→ Subscribe to the podcast: Blindspot Capital
→ Join the FemmeHealth Alliance Circle
The Hidden Risk Behind Standard Dosing Rules
As I followed the thread, I felt an eerie echo of something I lived long before AI ever entered clinical workflows.
During the Global Financial Crises of 2008, I remember standing on the trading floor of a large investment bank, watching my Bloomberg terminal across three screens glow red. There was a strange quiet in the room, the kind of silence that forms when people no longer trust what they’re seeing.
Diversification, the framework meant to protect portfolios, had shifted in real time. Financial assets that should have moved independently, assets that were designed to act as a counterbalance were suddenly moving in sync. It wasn’t that the concept of diversification was wrong. It worked…until it didn’t.
In 2008, diversification, as a risk management framework wasn’t fragile because it was flawed. It was fragile because it assumed stability across cases where stability did not exist.
It’s a reminder I have carried ever since: rules are only as safe as the exceptions they fail to encode. And the exceptions always matter most when the stakes are highest.
Why the Average Patient Doesn’t Exist in Real Biology
This is the same mistake medicine makes when it assumes uniform metabolism, symptoms, drug response, clearance, and risk.
The oxacillin incident of last week reveals this truth. Systems, whether financial or medical, are built on generalizations but human beings survive on nuance because biology isn’t uniform.
Clinicians recognized the oxacillin error immediately because medicine teaches both the general renal dosing rule and the nuanced truth that oxacillin doesn’t follow the general rule.
But AI models sometimes don’t learn nuance. They learn patterns. And they apply them broadly, confidently, and often invisibly to the average case.
The average patient is a myth. A convenience. A statistical compression of the messy, divergent, unpredictable truth of real human biology, and one of medicine’s most dangerous assumptions.
❤️ Enjoying this? If this post sparked something for you, click the ❤️ at the bottom. It helps more than you know and tells me you're reading.
Digoxin and the Cost of Assuming All Patients Are the Same
This would not be the first time generalized rules in medicine have masked dangerous exceptions.
More than 30 years ago, researchers discovered that digoxin, a widely used heart medication, behaved differently in women. Women given the same “standard dose” as men reached higher serum levels and suffered higher mortality rates.
The average trial data looked fine. The guideline looked fine. The general rule looked fine. But the exception, female physiology revealed the fatal oversight.
Digoxin didn’t harm women. The assumption that women and men respond the same way did.
Digoxin wasn’t the problem. The assumption of sameness was. And it has taken decades for the system to confront that truth.
Biological Variation Is the Missing Variable
I often think back to that market crash in 2008 and the realization that even elegant models can hide fragile foundations until stress exposes the missing variable.
I see echoes of that in medicine today. And the missing variable is biological variation; metabolism, hormones, sex differences, midlife physiology, immune divergence. This is variation the system has treated as noise because “the average” makes everything look stable. But “the average” hides the truth.
Whether it’s a market crash or a medication dose, the cost of ignoring variation shows up suddenly, systemically, and painfully.
Variation isn’t a flaw in biology. It’s its intelligence.
Why Precision and Personalization Will Drive the Next Decade of Health Innovation
So let me offer a prediction from someone who has lived in both worlds.
Precision is where value will aggregate in the next decade.
Not because personalization is fashionable, but because it offers a safer, more accurate, more efficient, and more economically sound way of practicing medicine.
Precision isn’t about tailoring care for comfort; it is about reducing risk, strengthening safety, improving efficacy, allocating capital intelligently, and building models that hold under stress.
Precision demands of us to move away from treating the average as truth and toward treating the individual as the unit of analysis. It shifts our attention from noise to signal, from surface patterns to underlying mechanisms, from rigid rules to the exceptions that expose their limits, from populations to people.
The oxacillin error was small, but the lesson it crystallizes is enormous. The next frontier of healthcare value sits where biology diverges, not where it conforms.
The Precision Premium
AI will get better. But only if we design it for the world as it actually is and not the world that averages pretend to describe.
The future of medicine and the future of health innovation will not belong to systems that perfect the rule but to the systems that notice the exception first.
Because the exception isn’t the edge case. The exception is the truth the rule has not yet learned to hold. And that is exactly where the next decade of science, care, and capital will be built.
Because biological variation isn’t noise. It’s information density. It’s alpha.
It’s the Precision Premium.
If this essay moved you, share or restack it.
This essay is part of my ongoing series on the billion-dollar blind spot in women’s health, wealth, and capital. You can subscribe to receive new essays every Sunday and join the conversation shaping the future of investing
To go deeper, pre-order my upcoming book The Billion Dollar Blind Spot to learn why women’s health is the future of healthcare investing.
Join Our Network
Are you building or backing credible, under-the-radar solutions in women’s health?
We want to hear from you. Reach out privately or reply to this post. We curate brands and breakthroughs that deserve broader attention in the women’s health ecosystem.
I write weekly at The Billion Dollar Blind Spot about capital, care, and the future of overlooked markets. If you are building, backing, or allocating in this space, I’d love to connect.
Disclaimer & Disclosure
This content is for informational and educational purposes only. It does not constitute financial, investment, legal, or medical advice, or an offer to buy or sell any securities. Opinions expressed are those of the author and may not reflect the views of affiliated organisations. Readers should seek professional advice tailored to their individual circumstances before making investment decisions. Investing involves risk, including potential loss of principal. Past performance does not guarantee future results.
References
Pea, F., Viale, P., Pavan, F., & Furlanut, M. (2007). Pharmacokinetic considerations for antimicrobial therapy in patients receiving renal replacement therapy. Clinical pharmacokinetics, 46(12), 997–1038. https://doi.org/10.2165/00003088-200746120-00003
Rathore SS, Wang Y, Krumholz HM. Sex-based differences in the effect of digoxin for the treatment of heart failure. N Engl J Med. 2002;346(18):1403-1411.
Ahmed A, Bertrand M, Love T E, et al. Serum digoxin concentration and outcomes in women with heart failure. Am Heart J. 2005;149(4):803-810.
Adams KF Jr, Patterson JH, Ghali JK, et al. Relationship of serum digoxin concentration to mortality in women with heart failure. Eur J Heart Fail. 2005;7(8):842-847.
This essay made me reflect on how often we are treated as averages, not individuals. I’m curious: what’s one moment when you realised you weren’t the ‘typical case’ the system assumed you were?
I’ll assert that precision fitness will become a thing too. Love this discussion about the exceptions to the norms—you write so eloquently about the gender gap in research!