AI in Medicine News — 2026-11-05 (Thu)

Overview

Regulation and safety moved most: Google published health AI evaluation frameworks (The Tech Buzz, Unite.AI), US states are split on regulating AI in medicine (Newsweek), and the FDA's approach to generative AI is contested (RAPS). In clinical settings, Mayo Clinic rolled out decision-support AI tools (Post Bulletin), and documentation and EHR integration drew several items (HIT Consultant, Security Boulevard). All items are headlines only, so the strength of evidence behind each result is mostly unstated.

Diagnostics and imaging

- A study finds Americans wary of AI in diagnosis and insurance decisions (BenefitsPRO). This is a survey-type finding on attitudes, not on diagnostic accuracy. - Medscape asks in a headline, "Microsoft AI vs Physicians: Who Makes the Better Diagnosis?" The headline gives no result (Medscape). - An AI tool helps pathology residents read prostate biopsies more accurately. The headline gives no accuracy figure or study stage (Bioengineer.org). - GE HealthCare (GEHC) adds lung and aortic stenosis AI tools to Vscan Air. The headline does not state regulatory status (AuntMinnie). - ASPNR 2027 will be held in Miami with expanded education and AI initiatives (Applied Radiology).

Clinical decision support

- Mayo Clinic Platform developed an AI-enabled clinical decision support solution (HCI Innovation Group). Mayo Clinic is rolling out decision-support AI tools for its clinicians (Post Bulletin). - Anesthesiologists argue that perioperative AI needs to put humans first (Bioengineer.org). - Epic Systems says nurses answer patient messages faster and gain time for care with Epic AI. This comes from the vendor's own release (Epic Systems). - Diagnostic Imaging asks whether a shift from passive AI users to active AI creators could help address radiology's workforce crisis (Diagnostic Imaging). 【Generative AI and medical documentation】 - athenahealth launched athenaPatient MCP to bring official patient records into Anthropic's Claude (HIT Consultant). - Security Boulevard describes how to integrate ambient listening AI with Epic, Oracle Health and legacy EHR systems without replacing them (Security Boulevard). Nerdbot covers how to choose an EHR with AI clinical documentation (Nerdbot). - Behavioral Health Business asks whether AI will help behavioral health catch up after EHR designs left it behind (Behavioral Health Business). - TailorMed secured majority private equity investment to expand AI-driven pharmacy access (HIT Consultant). 【Hospital operations and care delivery】 - AI promises to cut healthcare costs and improve efficiency, but cheaper treatment remains an unproven promise (The South First). Another item says the real cost of healthcare AI begins with the care pathway (Innovation News Network). - The NHS workforce plan is being reassessed due to AI advances, according to the health secretary (Pulse Today). - Pittsburgh experts weigh in on who is responsible when AI makes medical mistakes (TribLIVE.com). - Mount Sinai chose Sectra's AI-assisted radiology reporting (Unite.AI). 【Drug discovery and clinical trials】 None of the headlines states a development stage (preclinical, Phase I, etc.). - Some biotech startups are skipping AI for drug discovery (STAT). Drugmaker CEOs are not buying Silicon Valley's promise that AI will cure disease, per the WSJ headline (WSJ). - AI, automation and new assays power drug discovery (Newswise). - nPhase raised $30M to expand its AI-native clinical data platform (Fierce Biotech). - Xtalks is hosting a webinar on AI governance in clinical trial biometrics, a practical framework for small and mid-size organizations (Yahoo Finance). 【Medical device approval and regulation】 - Clearance: Neurotronics received FDA clearance for AI sleep analysis software (MobiHealthNews). - Policy: stakeholders offer competing views on the FDA's regulatory approach to generative AI (RAPS). A retrospective analysis finds FDA warning letters rarely mention AI (Bioengineer.org). - Draft guidance: a piece explains what Europe's draft Annex 22 asks of human oversight of AI (Bioprocess Online). - Commentary: a new AI architecture could help medtech firms stay ahead of AI-powered FDA reviews (Medical Device and Diagnostic Industry).

Safety, errors and bias

- Google unveiled a safety framework for medical AI development (The Tech Buzz) and published a multi-method framework for evaluating health AI (Unite.AI). - AI in healthcare raises health equity, algorithmic bias and disability risks (Beinsure). - States are split on regulating AI in medicine (Newsweek). - Felix launched a new AI assistant that brings in a patient's own doctor, in contrast to health AI that ends with "Talk to Your Doctor." (PR Newswire Canada) 【Research results (major journals)】 The headlines do not name the journals or study designs. - AI predicts patient survival from cardiac MRI scans better than clinicians (healthcare-in-europe.com). - AI chatbots worsen psychiatric symptoms in a study (Chosun Ilbo). - Medical education must adapt to AI (HealthLeaders Media). - Google shared its framework for building safer and more helpful AI for health (blog.google). - An overview piece covers what AI in medicine actually does today (Pasquale Pillitteri).

Developments in Japan

- Nikkei Medical asks whether AI can be entrusted with work permitted only to physicians (Nikkei Medical). Nikkei reports that Japan's Medical Practitioners Act stands in the way of the day AI is recognized as a physician (Nikkei). - Gunma University Hospital is using generative AI to draft medical documents so staff can focus on patient care (Joumou Shimbun). - ThinkIT describes how Kaggle helped one engineer move from embedded systems to image AI to medical AI (ThinkIT). - Forbes Japan covers a "medical revolution" from combining AI and quantum computing, with a claim of breaking through limits in drug discovery. The headline gives no evidence stage (Forbes Japan).

Statements by key figures

No notable developments.