https://omeratli.com Writing on healthcare AI, emergency medicine, clinical safety, and medical evidence. en-GB https://omeratli.com/writing/screening-is-not-always-a-gift https://omeratli.com/writing/screening-is-not-always-a-gift Fri, 05 Jun 2026 00:00:00 GMT "Early detection saves lives" is the most intuitive sentence in medicine — and one of the easiest to prove without proving anything at all. https://omeratli.com/writing/the-confident-wrong-answer https://omeratli.com/writing/the-confident-wrong-answer Fri, 05 Jun 2026 00:00:00 GMT Traditional clinical software fails in ways you can anticipate. AI fails differently — fluently, confidently, and most dangerously when it is wrong. Safety thinking has to change to match. https://omeratli.com/writing/who-is-your-clinical-safety-officer https://omeratli.com/writing/who-is-your-clinical-safety-officer Fri, 05 Jun 2026 00:00:00 GMT Many digital health products have a named clinical safety officer and no real one. The gap between the title and the function is where safety quietly stops happening. https://omeratli.com/writing/compensation-hides-the-crash https://omeratli.com/writing/compensation-hides-the-crash Thu, 04 Jun 2026 00:00:00 GMT The body is built to mask its own emergencies — and the numbers are the last thing to tell you it has run out of road. https://omeratli.com/writing/hazard-is-not-risk https://omeratli.com/writing/hazard-is-not-risk Thu, 04 Jun 2026 00:00:00 GMT Two words that digital health teams use interchangeably mean very different things. The confusion isn't pedantic — it's the mechanism by which real harm gets reasoned away. https://omeratli.com/writing/the-safety-case-is-an-argument https://omeratli.com/writing/the-safety-case-is-an-argument Thu, 04 Jun 2026 00:00:00 GMT Most "safety cases" in digital health are collections of documents that prove activity occurred. A real one is a reasoned, falsifiable argument that a specific system is acceptably safe — and the difference is everything. https://omeratli.com/writing/quiet-failures-are-the-dangerous-ones https://omeratli.com/writing/quiet-failures-are-the-dangerous-ones Wed, 03 Jun 2026 00:00:00 GMT The failures that hurt patients in digital health are almost never the crashes. They're the silences — the result that never arrives and never announces it didn't. https://omeratli.com/writing/the-pilot-that-never-ends https://omeratli.com/writing/the-pilot-that-never-ends Wed, 03 Jun 2026 00:00:00 GMT The most common outcome of a healthcare AI pilot is not success or failure. It's another pilot. https://omeratli.com/writing/write-medicine-in-plain-english https://omeratli.com/writing/write-medicine-in-plain-english Wed, 03 Jun 2026 00:00:00 GMT Plain English in medicine is not simpler writing — it is the writer doing the work so the reader doesn't have to. https://omeratli.com/writing/time-is-a-diagnostic-test https://omeratli.com/writing/time-is-a-diagnostic-test Tue, 02 Jun 2026 00:00:00 GMT The most informative investigation for the undifferentiated abdomen is sometimes not in the radiology department. It is four hours and a second examination. https://omeratli.com/writing/why-case-reports-still-matter https://omeratli.com/writing/why-case-reports-still-matter Tue, 02 Jun 2026 00:00:00 GMT It sits at the bottom of the evidence pyramid because the pyramid measures the wrong thing. https://omeratli.com/writing/ai-scribes-are-not-the-endgame https://omeratli.com/writing/ai-scribes-are-not-the-endgame Mon, 01 Jun 2026 00:00:00 GMT AI scribes solve a real documentation problem. But calling them co-pilots confuses transcription with clinical reasoning — and the gap matters. https://omeratli.com/writing/automation-bias-has-a-bedside https://omeratli.com/writing/automation-bias-has-a-bedside Mon, 01 Jun 2026 00:00:00 GMT The dangerous failure of clinical AI is rarely the model being wrong — it's the clinician agreeing with it anyway. https://omeratli.com/writing/shadow-ai-is-already-in-the-hospital https://omeratli.com/writing/shadow-ai-is-already-in-the-hospital Mon, 01 Jun 2026 00:00:00 GMT The most widely used clinical AI in any hospital today was never procured, never assessed, and never appears on a single risk log. It is in the staff's pockets. https://omeratli.com/writing/the-update-is-a-clinical-intervention https://omeratli.com/writing/the-update-is-a-clinical-intervention Mon, 01 Jun 2026 00:00:00 GMT When a medicine changes, a process begins. When clinical software changes, a release note nobody reads gets filed — and the same patients are affected anyway. https://omeratli.com/writing/the-limitations-section-is-the-honest-bit https://omeratli.com/writing/the-limitations-section-is-the-honest-bit Sun, 31 May 2026 00:00:00 GMT The abstract is written for the press release. The limitations are written for the three people who can end the authors' careers. https://omeratli.com/writing/conflicts-of-interest-beyond-the-disclosure-slide https://omeratli.com/writing/conflicts-of-interest-beyond-the-disclosure-slide Sat, 30 May 2026 00:00:00 GMT The most effective conflicts of interest produce papers that pass every integrity check — because the money was spent before the study began. https://omeratli.com/writing/the-second-visit-rule https://omeratli.com/writing/the-second-visit-rule Sat, 30 May 2026 00:00:00 GMT Re-attendance is one of the highest-yield danger signals in medicine, and it is the one most reliably read as the opposite. https://omeratli.com/writing/what-digital-health-should-steal-from-aviation https://omeratli.com/writing/what-digital-health-should-steal-from-aviation Sat, 30 May 2026 00:00:00 GMT "Medicine should be more like aviation" is the most repeated and least examined claim in patient safety. Some of it is gold; most of the laminated cargo is not. https://omeratli.com/writing/the-surrogate-endpoint-problem https://omeratli.com/writing/the-surrogate-endpoint-problem Fri, 29 May 2026 00:00:00 GMT A trial can move the number it measures and miss the thing it was meant to fix entirely — and the history of medicine is built on the rubble of confusing the two. https://omeratli.com/writing/what-clinical-ai-evals-actually-measure https://omeratli.com/writing/what-clinical-ai-evals-actually-measure Fri, 29 May 2026 00:00:00 GMT A model that aces the membership exam has proven one thing: that the exam was automatable. Nobody's shift got safer. https://omeratli.com/writing/alert-fatigue-is-a-design-failure https://omeratli.com/writing/alert-fatigue-is-a-design-failure Thu, 28 May 2026 00:00:00 GMT When a clinician clicks through a warning without reading it, the system has not been ignored. It has been understood. https://omeratli.com/writing/gestalt-is-data https://omeratli.com/writing/gestalt-is-data Thu, 28 May 2026 00:00:00 GMT "The patient looks unwell" sounds like the least scientific sentence in medicine. It is closer to the most compressed one. https://omeratli.com/writing/worst-first-thinking https://omeratli.com/writing/worst-first-thinking Thu, 28 May 2026 00:00:00 GMT The emergency physician is not trying to work out what you have. They are trying to work out what you'd better not have. https://omeratli.com/writing/why-doctors-need-to-understand-ai https://omeratli.com/writing/why-doctors-need-to-understand-ai Wed, 27 May 2026 00:00:00 GMT AI now operates on clinical language and reasoning, and the defaults are being set now — understanding it is how doctors get to set the terms instead of inheriting them. https://omeratli.com/writing/data-quality-is-a-patient-safety-issue https://omeratli.com/writing/data-quality-is-a-patient-safety-issue Tue, 26 May 2026 00:00:00 GMT A wrong allergy entered as a precaution a decade ago is still quietly removing the best drug from every decision made since — and nobody filed that as harm. https://omeratli.com/writing/ground-truth-is-shakier-than-it-looks https://omeratli.com/writing/ground-truth-is-shakier-than-it-looks Tue, 26 May 2026 00:00:00 GMT Every medical AI accuracy figure is measured against a label, and the label is mostly opinion plus administration, frozen at an arbitrary moment. https://omeratli.com/writing/relative-risk-sells-absolute-risk-informs https://omeratli.com/writing/relative-risk-sells-absolute-risk-informs Tue, 26 May 2026 00:00:00 GMT The same finding, framed two ways, tells two different stories — and only one of them is honest with you. https://omeratli.com/writing/the-handover-is-the-most-dangerous-ten-minutes https://omeratli.com/writing/the-handover-is-the-most-dangerous-ten-minutes Mon, 25 May 2026 00:00:00 GMT Medicine's riskiest moments aren't the procedures. They're the transitions — and the most lossy transition of all happens at every shift change. https://omeratli.com/writing/the-workaround-is-the-real-workflow https://omeratli.com/writing/the-workaround-is-the-real-workflow Mon, 25 May 2026 00:00:00 GMT Every workaround on a ward is a field report from the exact spot where the designed system fails the work. https://omeratli.com/writing/the-demo-is-not-the-deployment https://omeratli.com/writing/the-demo-is-not-the-deployment Sun, 24 May 2026 00:00:00 GMT Healthcare AI is bought on its best day and used on its worst, and the distance between the two is where most of the failures live. https://omeratli.com/writing/the-emergency-department-test-for-medical-ai https://omeratli.com/writing/the-emergency-department-test-for-medical-ai Sun, 24 May 2026 00:00:00 GMT A five-question test for medical AI: how it handles noise, missing data, interruption, the unselected tail, and who answers when it's wrong. https://omeratli.com/writing/the-missing-denominator https://omeratli.com/writing/the-missing-denominator Sun, 24 May 2026 00:00:00 GMT Every triumphant AI success story is a number on top of a fraction whose bottom half nobody will print. https://omeratli.com/writing/chest-pain-is-a-trapdoor https://omeratli.com/writing/chest-pain-is-a-trapdoor Sat, 23 May 2026 00:00:00 GMT Chest pain is one label over conditions with opposite stakes; separating the lethal from the benign is consequence-weighted exclusion, not puzzle-solving. https://omeratli.com/writing/how-to-read-a-medical-paper https://omeratli.com/writing/how-to-read-a-medical-paper Fri, 22 May 2026 00:00:00 GMT Read methods first, numbers second, the authors' opinions third, and the abstract last — because the abstract is the paper's advertisement, not its evidence. https://omeratli.com/writing/the-healthy-user-problem https://omeratli.com/writing/the-healthy-user-problem Fri, 22 May 2026 00:00:00 GMT Most "X linked to longer life" stories aren't measuring X. They're measuring the kind of person who does X. https://omeratli.com/writing/triage-is-the-operating-system-of-medicine https://omeratli.com/writing/triage-is-the-operating-system-of-medicine Fri, 22 May 2026 00:00:00 GMT Triage — comparative prioritisation under scarcity — is medicine's core function and healthcare AI's highest-leverage, highest-risk, most neglected problem. https://omeratli.com/writing/the-return-of-the-medical-generalist https://omeratli.com/writing/the-return-of-the-medical-generalist Thu, 21 May 2026 00:00:00 GMT Specialisation solved a storage problem AI now dissolves; the scarce skill becomes integrating care across domains for multimorbid patients. https://omeratli.com/writing/why-doctors-miss-pulmonary-embolism https://omeratli.com/writing/why-doctors-miss-pulmonary-embolism Thu, 21 May 2026 00:00:00 GMT Pulmonary embolism is missed through ordinary reasoning mechanisms, not ignorance — a tour of how diagnostic error actually happens anywhere. https://omeratli.com/writing/clinical-safety-is-not-a-checkbox https://omeratli.com/writing/clinical-safety-is-not-a-checkbox Wed, 20 May 2026 00:00:00 GMT The test of safety work is whether it changed the product. If the hazard log was written and nothing was redesigned, what happened was compliance theatre. https://omeratli.com/writing/discharge-is-a-clinical-decision-too https://omeratli.com/writing/discharge-is-a-clinical-decision-too Wed, 20 May 2026 00:00:00 GMT We drill admission criteria for years and learn to send people home by osmosis. That is exactly backwards. https://omeratli.com/writing/the-nhs-tone https://omeratli.com/writing/the-nhs-tone Wed, 20 May 2026 00:00:00 GMT The flat NHS register is disciplined risk communication: it keeps conditionals attached, sizes uncertainty, and saves its dynamic range for what is urgent. https://omeratli.com/writing/the-problem-with-symptom-checkers https://omeratli.com/writing/the-problem-with-symptom-checkers Wed, 20 May 2026 00:00:00 GMT The symptom checker trilemma is structural, not an execution problem: no correct threshold exists at the information level the product operates on. https://omeratli.com/writing/patients-want-interpretation-not-information https://omeratli.com/writing/patients-want-interpretation-not-information Tue, 19 May 2026 00:00:00 GMT Medicine solved information abundance long ago; the real shortage is interpretation — accountable weighting of what a condition means for this person. https://omeratli.com/writing/blame-is-not-an-analysis https://omeratli.com/writing/blame-is-not-an-analysis Mon, 18 May 2026 00:00:00 GMT "Human error" is the most popular conclusion in healthcare, and one of the least useful — it names who was holding the failure, not what built it. https://omeratli.com/writing/medical-content-review-accuracy-is-not-enough https://omeratli.com/writing/medical-content-review-accuracy-is-not-enough Mon, 18 May 2026 00:00:00 GMT Reviewing health content for safety is a different discipline from fact-checking it, and AI has made the whole-page failure modes cheap to produce at scale. https://omeratli.com/writing/why-fatigue-is-medicines-hardest-symptom https://omeratli.com/writing/why-fatigue-is-medicines-hardest-symptom Mon, 18 May 2026 00:00:00 GMT Fatigue defeats the diagnostic machinery built for specificity; the work lives in disciplined history-taking and honestly managed uncertainty, not the lab. https://omeratli.com/writing/normal-blood-tests-are-not-the-whole-story https://omeratli.com/writing/normal-blood-tests-are-not-the-whole-story Sun, 17 May 2026 00:00:00 GMT A normal blood panel is not a verdict but a set of narrow, time-stamped, question-shaped answers, meaningful only against the story behind them. https://omeratli.com/writing/probably-anxiety-the-most-expensive-phrase https://omeratli.com/writing/probably-anxiety-the-most-expensive-phrase Sun, 17 May 2026 00:00:00 GMT 'Probably anxiety' is dangerous because it's usually right — cheaply, self-sealingly, and most often at the expense of the patients the system hears least. https://omeratli.com/writing/the-future-doctor-clinician-editor-systems-thinker https://omeratli.com/writing/the-future-doctor-clinician-editor-systems-thinker Sat, 16 May 2026 00:00:00 GMT Tomorrow's doctor stays a clinician but adds two skills training ignores: editing machine-drafted clinical text and reasoning about the systems that now decide. https://omeratli.com/writing/what-healthcare-startups-misunderstand-about-doctors https://omeratli.com/writing/what-healthcare-startups-misunderstand-about-doctors Sat, 16 May 2026 00:00:00 GMT Six specific ways healthtech product teams build around a wrong model of how clinicians actually work — and what to do instead. https://omeratli.com/writing/who-answers-when-the-model-is-wrong https://omeratli.com/writing/who-answers-when-the-model-is-wrong Sat, 16 May 2026 00:00:00 GMT When an AI-influenced clinical decision goes bad, the accountability lands on the one person in the chain with the least insight into how the system works — and that arrangement is living on borrowed time. https://omeratli.com/writing/resource-limited-emergency-medicine-risk https://omeratli.com/writing/resource-limited-emergency-medicine-risk Fri, 15 May 2026 00:00:00 GMT Practising emergency medicine without abundant resources strips clinical reasoning to its load-bearing structure — and reveals what decision-support technology still misses. https://omeratli.com/writing/why-vague-symptoms-are-where-medicine-gets-dangerous https://omeratli.com/writing/why-vague-symptoms-are-where-medicine-gets-dangerous Fri, 15 May 2026 00:00:00 GMT Vague presentations defeat protocols, degrade triage, and exploit pattern recognition's blind spots. How emergency medicine actually reasons under ambiguity.