Physician in the Loop

The White House wants one national AI rulebook instead of fifty. Physicians should accept it only if it keeps the states' rules on AI in medicine or protects patients at least as well.

Read the special topic and its sources: https://physicianintheloop.org/specials/fifty-rulebooks-for-now/

The voice on this episode is generated by AI. It reads a special topic from Physician in the Loop as written; every fact in it links to its source.

Software meant to save medicine time is sending its bills downstream, to the nurses who recheck it and the hospitals that host it. Whoever profits from it should pay for the checking.

Read the letter and its sources: https://physicianintheloop.org/letters/2026-10-19/

The voice on this episode is generated by AI. It reads the week's Friday letter at Physician in the Loop as written; every fact in the letter links to its source.

Microsoft diagnostic AI beats generalist physicians on hard cases, study says

The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.

Read the post and its sources

Get the free email briefing

Microsoft's multi-agent diagnostic system reached the correct diagnosis in 83.2% of complex New England Journal of Medicine cases, against 36.1% for generalist physicians, in a benchmark study Science published online Oct. 22.

The physicians in the comparison were primary care physicians and general internists working alone with reference materials, and no specialists or physician teams took part, TechCrunch reported. The authors described the results as coming from a controlled benchmark, and Microsoft said it plans a prospective study at Beth Israel Deaconess Medical Center in Boston in which patients interact with the system directly.

Elsewhere, the director of the Food and Drug Administration's Digital Health Center of Excellence said Oct. 20 that draft guidance on generative AI devices will come in the coming months, MedTech Dive reported. Sutter Health said more than 6,000 of its clinicians have drafted more than 5 million notes with Abridge's ambient AI, and the FDA approved Edwards Lifesciences' Pascal Precision system for transcatheter repair of the tricuspid valve.

In this episode

• Microsoft's AI diagnostic panel solves 83.2% of hard NEJM cases vs. 36.1% for generalist physicians, Science paper reports
• FDA digital health chief says draft generative AI device guidance is coming; device center lost about 22% of its staff
• Sutter Health says more than 6,000 clinicians have drafted 5 million notes with Abridge's ambient AI
• Google Research's personal health agent ranks first for 46.7% of 45 end users vs 24.4% for a single-agent baseline in a rating study
• FDA approves Edwards' Pascal Precision system for transcatheter repair of severe tricuspid regurgitation
• Hims & Hers asks court to dismiss FTC suit over sharing customer data with ad platforms and subscription billing
• Baba raises $40 million for care coordination that pairs AI with licensed clinicians for Medicare and Medicaid patients

AI that drafts replies to patients has not reliably saved doctors time. Health systems should first use AI to sort the inbox, and draft clinical replies only where audits show a gain.

Read the special topic and its sources: https://physicianintheloop.org/specials/route-dont-write/

The voice on this episode is generated by AI. It reads a special topic from Physician in the Loop as written; every fact in it links to its source.

Epic lets its AI assistants reach into other companies' apps

The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.

Read the post and its sources

Get the free email briefing

Epic Systems on Oct. 21 introduced Connectors, which link outside developers' apps to its electronic health record so Epic's AI assistants can pull information from those apps and complete tasks in them.

Health care organizations can find developer-built Connectors in Epic's Showroom marketplace starting in March 2027 and can review what each one reads and changes before switching it on, the company said. Epic plans to begin testing with some health system partners by the end of 2026 and will charge developers to distribute Connectors, Fierce Healthcare reported. The announcement came at Epic's Open@Epic developer conference, where an Epic executive also said consumer chatbots that pull patient records are adding traffic and cost for health systems.

Elsewhere, the American Hospital Association asked the Food and Drug Administration to make AI developers primarily responsible for monitoring generative AI devices, and five large employers backed extending electronic prior authorization to self-funded health plans, the Centers for Medicare & Medicaid Services said.

In this episode

• Epic opens its AI assistants to third-party apps through Connectors, with marketplace listings from March 2027
• Consumer chatbots pulling patient records add traffic and cost for Epic customers; Epic adds event alerts to cut repeated data requests
• Hospital association asks FDA to make developers primarily responsible for monitoring generative AI devices after they are sold
• Amazon, Google, Meta, Oracle and Axon back extending electronic prior authorization to self-funded employer plans
• Nebraska Medicine shelves its in-house AI cancer registry tool after it reaches a little over 80% accuracy against a 95% bar
• FDA clears Elucid's FFR-IQ, which estimates fractional flow reserve from coronary CT angiography using plaque features
• Doximity adds UpToDate answers to its Ask clinical AI tool for clinicians at health systems that use both products
• FDA clears NeuroXT's NEXT-Cue, AI software for assessing whether mild cognitive impairment will progress to Alzheimer's dementia

Chatbot-linked psychosis appears rare, but hospital counts depend on what clinicians happened to write down. Physicians should ask about chatbots in every new psychosis or mania, and record the answer the same way.

Read the special topic and its sources: https://physicianintheloop.org/specials/count-the-cases/

The voice on this episode is generated by AI. It reads a special topic from Physician in the Loop as written; every fact in it links to its source.

AI model estimates cancer patients' survival from a single face photo

The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.

Read the post and its sources

Get the free email briefing

Mass General Brigham researchers reported Oct. 19 that an AI model trained on more than 40 million face photographs predicted survival in cancer patients, sorting them into risk groups whose mortality differed more than threefold.

The study, published in The Lancet Digital Health, tested the model in two independent cancer cohorts, and the authors said a face photo could help with risk stratification when other biomarkers are unavailable. No prospective clinical use has been reported. Separately, the Food and Drug Administration finalized guidance on how drugmakers label apps sold for use with prescription drugs, Radical Numerics said phages designed by its AI model were given to a patient with a drug-resistant infection, and Intuitive reported that da Vinci procedures grew about 14% in the third quarter.

In this episode

• AI model trained on 40 million face photos sorts cancer patients by survival risk, Lancet Digital Health study finds
• FDA finalizes guidance on when apps sold with prescription drugs belong in the drug's prescribing information
• Radical Numerics says phages designed by its AI model were given to a patient with an antibiotic-resistant infection
• Intuitive says da Vinci procedures grew about 14% in the third quarter and places 546 more systems
• Home and virtual supportive oncology program tied to 10.4% fewer admissions among 5,432 Rhode Island cancer patients

When ransomware locked hospitals' computers, more of the Medicare patients already in their beds died. As care depends on more software, AI included, keeping that software safe must come before speed or savings.

Read the special topic and its sources: https://physicianintheloop.org/specials/breach-is-a-clinical-event/

The voice on this episode is generated by AI. It reads a special topic from Physician in the Loop as written; every fact in it links to its source.

Doximity and OpenEvidence settle dueling lawsuits over physician AI tools

The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.

Read the post and its sources

Get the free email briefing

Doximity and OpenEvidence dismissed all claims in their federal lawsuit over physician AI tools, ending a case in which OpenEvidence accused Doximity staff of posing as physicians, Fierce Healthcare reported Oct. 19.

The claims and counterclaims were dismissed with prejudice, and each company bears its own costs and fees; the joint notice disclosed no other terms. Separately, Oxford researchers reported a labor-onset model that pairs clinical risk factors with computerized fetal heart tracing, and a BMJ review found that trials of large language model tools often leave out model versions, code and harms. The day brought few other developments.

In this episode

• Doximity and OpenEvidence dismiss all claims in lawsuit that accused Doximity staff of posing as physicians to probe its AI
• Model pairing clinical risk factors with 20 minutes of computerized fetal heart tracing sorts labor risk across 147,350 births, Lancet Digital Health study finds
• Trials of large language model tools often omit model versions, code and harms reporting, BMJ review of 461 trial records finds
• XRlabs raises $8.4 million for operating-room software that merges live surgical video, navigation and imaging; it has no FDA clearance
• Siemens Healthineers and Paradromics build an imaging workflow for implanting brain-computer interfaces, starting with speech

When a patient links a medical record to a chatbot, HIPAA stays behind. The protections should go with the record; until Congress acts, physicians should tell patients plainly what they give up.

Read the special topic and its sources: https://physicianintheloop.org/specials/chart-leaves-the-building/

The voice on this episode is generated by AI. It reads a special topic from Physician in the Loop as written; every fact in it links to its source.