The library
The AI in healthcare library
Everything AIMOCS publishes, open to read. Guides, a glossary, statistics trackers and the briefing archive.
Start here
The clinical AI evaluation kit
The questions to ask a vendor, a checklist reconciling the published evaluation frameworks, the red flags in a reported performance figure, and a governance committee charter with an intake form.
What this is
How the library is written
Every piece carries the same three things: numbered primary sources you can open yourself, the date it was last updated, and the name of the person who reviewed it where a person did. Nothing is bylined to an invented author.
Where the evidence is thin, the piece says so instead of rounding up, and where a figure has been superseded the tracker is corrected in the open with the date on it. Three pieces a week is the ceiling. The editorial policy sets out how sourcing, review and corrections are handled.
Guides
Guides: evaluating, deploying and governing
The long pieces. How to read a validation study that was written to be persuasive, what the FDA has actually cleared and what a Predetermined Change Control Plan lets a vendor change afterwards, how ambient documentation behaves once it is in a real clinic, and what a governance committee has to decide that no regulator decides for it. Six collections, each with its own index.
Evaluation 13Ambient AI 11Agentic AI 10Regulation 16Specialties 12Comparisons 11
The best AI in healthcare communities, societies, and networks in 2026
Twelve places where the people doing AI in healthcare actually gather — CHAI, Health AI Partnership, AMIA, SIIM, AIME, DiMe, AAIH, the FHIR chat, Medblocks, Health Tech Nerds, Out-Of-Pocket, and AIMOCS — compared on who they serve, what membership really looks like, and every published fee, each figure from the organization's own pages. As of 12 August 2026.
How to get into AI in healthcare: a clinician's guide
The realistic map for a physician, nurse leader, or researcher entering the field in 2026: five destinations, what each actually requires, where formal credentials matter and where they are myth, a self-directed 90-day reading path, and when a community shortens all of it. Every number sourced and dated. As of 12 August 2026.
The best AI in healthcare courses in 2026, compared
Johns Hopkins, Harvard, Stanford, MIT Sloan, and AIMOCS, weighed on what each programme publishes about itself: how current the syllabus is, how deep the generative and agentic coverage goes, what you build, what continues after the certificate, and who carries accredited credit. Every cell sourced and dated. As of 1 August 2026.
AI scribe vendor landscape 2026
A dated, neutral map of the ambient AI scribe field — what each vendor documents, what it has disclosed raising, and which independent evaluations actually exist — with every cell tied to a source and no leader named. As of July 2026.
AI scribes: what the evidence actually shows
Ambient scribes reached millions of uses before the first randomized trial reported a result. This guide sets the deployment scale and the vendor efficiency claims against the peer-reviewed record — how few evaluations meet real-world-evidence criteria, and what the studies that do exist actually found. As of July 2026.
Ambient AI ROI calculator: a transparent model
A fully worked return model for ambient AI scribes — every input, formula, and example laid out and drawn from published figures — expressed in clinician-time and visit-capacity terms. A model to test locally, never a promise. As of July 2026.
Glossary
Glossary: the vocabulary, defined once
Terms defined the way a clinician needs them defined: what the word means, what it is routinely used to mean instead, and where the definition comes from. Short entries, each with its sources. These are the pages to send to a colleague who is about to sit in a vendor meeting.
Clinical LLM
What a clinical LLM is, how general models are adapted for healthcare work, what benchmark scores do and do not prove, and where regulators have drawn the lines. As of July 2026.
Federated learning in healthcare
What federated learning means when the data is patient data — how institutions train a shared model without moving records, what it has already delivered across dozens of hospitals, and why it is a privacy tool rather than a privacy guarantee. As of September 2026.
Agentic AI
What agentic AI means in healthcare — the working definition from the peer-reviewed literature, how these systems differ from chatbots, and what benchmarks show they can and cannot yet do. As of July 2026.
AI hallucination in clinical contexts
What it means when a clinical AI system fabricates content — the measured rates, the severity splits, and the failure modes specific to healthcare. As of July 2026.
Algorithmovigilance
The discipline of watching clinical algorithms after deployment — where the term came from, the failures that made it necessary, and the systems and rules now operationalizing it. As of July 2026.
Ambient AI scribe
What an ambient AI scribe is, how the capture-to-signed-note pipeline works, and what the largest deployments have measured — time saved, burnout moved, and the failure modes that keep clinicians reviewing every note. As of July 2026.
Statistics
Statistics: running trackers, not one-off posts
Adoption, device clearances, published evidence, funding. Each figure is normalised by what it actually counts — a number about “AI-enabled devices” and a number about “clinical AI in daily use” are not the same number — and carries the date it was true along with its primary source. Trackers are revisited on a schedule rather than left to rot.
LLM medical benchmark results tracker
A running read of how large language models score on the leading clinical benchmarks — MedHELM, HealthBench, and the new HealthBench Professional — set beside the peer-reviewed critique of what those scores do, and do not, tell you about bedside performance. As of August 2026.
Healthcare AI funding and M&A tracker
What the three primary trackers — Rock Health, CB Insights, and Silicon Valley Bank — actually report about digital-health and healthcare-AI funding across 2024–2026, why their credible totals differ by 3x, and the disclosed deals behind the headlines. As of July 2026.
Physician attitudes to AI: the survey tracker
Three waves of the AMA's Augmented Intelligence survey — 2023, 2024, and 2026 — placed on one time series, re-read with a consistent denominator, and cross-checked against an independent physician survey. Use, enthusiasm, concern, and the training gap, each with its date. As of July 2026.
Global AI in health regulation tracker
A living status table of the rules governing AI in healthcare — what is in force, what is proposed, and the exact effective dates — across the EU, the US FDA, the UK MHRA, the WHO, and other jurisdictions, each row tied to a primary regulator source. As of July 2026.
AI in healthcare statistics (2026)
A primary-sourced dashboard of where AI in healthcare actually stands — adoption, clinical evidence, regulation, and investment — with every figure dated and normalized by what it really counts. The hub for our statistics cluster. As of August 2026.
AI scribe adoption statistics
A running tally of how far ambient AI documentation has actually spread in healthcare — the deployments, the measured effects on clinician time and burnout, and the accuracy questions that stay open. As of July 2026.
Briefing
Briefing: one paper a week, read closely
What the paper measured, how well it measured it, and whether it should change anything. The archive stays open to everyone. The current issue goes out by email to members.
Briefing 001 — A generative AI copilot meets a hard patient endpoint
The first issue of the AIMOCS Briefing reads one study closely: a pragmatic, cluster-randomized trial that put a generative AI clinical copilot in front of clinicians treating nearly 10,000 patients in Kenya, then measured whether the patients did better. The notes got better. Within 14 days, the patients did the same either way — and that honest null is the most useful thing.
Briefing 002 — The largest ambient-scribe study yet counts the minutes
Ambient scribes are the fastest-spreading AI purchase in care delivery, and the pitch is time. Issue 002 reads the largest multisite cohort to date — 8,581 clinicians across five US academic health systems — and finds the time is real but modest: sixteen minutes of documentation per eight scheduled patient-hours. The after-hours record work, the part the sales pitch calls "your evenings back," did not move.
Briefing 003 — When the AI is switched off: the first deskilling signal
Four Polish endoscopy centres turned on AI polyp detection at the end of 2021. Issue 003 reads the study that looked sideways at what happened next: on procedures done without the software, experienced endoscopists detected adenomas in 22.4% of colonoscopies, against 28.4% in the months before AI arrived. It is observational, and it is the first real-world, patient-relevant deskilling signal on record — which makes reading its limits as important as reading its headline.
Keep up
The library is open. The weekly email is how people keep up with it.
One paper a week, read closely, by email. Joining also opens the paper feed and the member discussions, where questions get answered.
One email address