AI in Medicine News — 2026-11-07 (Sat)

Overview

Regulation and devices moved most: FDA clearance for an AI-powered intravascular ultrasound (IVUS) imaging platform, an FDA nod for an AI wearable arrhythmia analysis, a US Senate bill on expanding AI in healthcare, and Utah letting AI prescribe medicines (Radiology Business, Vascular News, Medical Product Outsourcing, Medical Buyer, Health and Me). In drug discovery, Roche agreed a deal with Chai Discovery, while scientists are wary of hype around Anthropic's biotech claims (AllSci, FirstWord Pharma, New York Post). Operations and safety coverage centered on the cost of deployment, trust among clinicians, and gaps between benchmarks and real workflows (Bioengineer.org, Telehealth.org).

Diagnostics and imaging

- A CCTA (coronary CT angiography) study using AI-powered plaque analysis found coronary plaque in 87 percent of retired NFL players. The headline describes a single study, not a validated clinical standard (Diagnostic Imaging). - A meta-analysis assessed AI accuracy in detecting pediatric appendicular fractures. The headline gives no accuracy figures, so the level reached is not stated (Bioengineer.org). - Academic medical centers will present new AI-ECG research on Viz.ai's Cardio Suite at AHA 2026. This is a conference presentation, and no results are given in the headline (FinancialContent). - Te Whatu Ora is exploring AI for surgery preparation and radiology referral. The headline describes exploration only (Healthcare IT News).

Clinical decision support

- Mayo Clinic unveiled a new AI-powered clinical decision support tool, described as a way to "individualize" patient care. The headlines give no performance or validation data (HealthExec). - Clinicians in the cardiothoracic ICU say AI must prove itself before they will trust it (Bioengineer.org). - Seattle Fire's AI 911 triage was rolled out with no public review, according to the city's privacy chief (Hoodline). - A piece titled "Can We Be Outsmarted By AI?" appeared. The headline alone gives no further content (HMP Global Learning Network). 【Generative AI and clinical documentation】 - A MarketScale piece argues that legacy EHR data decides whether healthcare AI pays off. This is an opinion-style headline, not a measured result (MarketScale). 【Hospital operations and care delivery】 - Digital pathology and AI efforts in Saudi Arabia face cost, workforce and data hurdles (Bioengineer.org). - In India, AI is finding its way into hospitals, but lower treatment costs remain unproven (Medical Buyer). - The NHS is trialing an AI phone assistant to monitor patients with long-term illness. This is a trial, and no outcomes are reported in the headline (University of York). - Illinois hospitals ran a drill for an AI-driven cyber attack (KMOV). - Charities are calling for safeguards for AI-generated health information (Digital Health). 【Drug discovery and clinical trials】 - Roche will use Chai Discovery's AI models in its drug discovery programs. The headlines call it a pact and part of Chai's dealmaking streak, and give no development stage (AllSci, FirstWord Pharma). - Inductive placed first in the OpenADMET CYP Inhibition Challenge, its fourth consecutive AI drug discovery benchmark win. This is a benchmark result, not a clinical-stage result (PR Newswire). - Anthropic is making biotech claims, but scientists are wary of hype around "curing cancer". No development stage is given in the headline (New York Post). 【Medical device approvals and regulation】 - Clearance: Evident Vascular received US FDA clearance for its high-fidelity AI-powered IVUS imaging platform (Vascular News, Radiology Business). - FDA nod (type not specified in the headline): PulseAI for AI wearable arrhythmia analysis (Medical Product Outsourcing). - Proposed rule: Treasury's proposed outbound investment rule adds biotech and AI prohibitions (Wiley Law). - Bill: a US Senate bill pushes for expanding the use of AI in healthcare (Medical Buyer). - A MedCity News piece argues that "The Big Challenge for AI in Healthcare Comes After Approval" (MedCity News). - Not medical: the EPA inspector general will audit the agency's own AI use (Chemical & Engineering News).

Safety, errors and bias

- Utah became the first US state to let AI prescribe medicines, and the coverage raises patient safety concerns (Health and Me). - A study compares medical AI benchmarks with real clinical workflows (Telehealth.org). - Medscape asks whether delaying healthcare AI could put patients at risk. The headline poses a question and states no finding (Medscape). 【Research findings (major journals)】 - A JAMA study of 952,008 clinicians examines what physicians are asking AI. The headline gives no findings (Passive Income MD). - A new study warns of dangers behind AI chatbots and "concerning" health advice from private clinics (Medical Xpress). - AI measures heart function in seconds, while doctors take minutes. The headline gives no accuracy data or study design (News-Medical). - A new review maps how real-time operating room models for surgical AI are built. This is a review, not a clinical trial (Bioengineer.org).

Developments in Japan

- Igaku-Shoin's newspaper reports a plan to implement trustworthy AI in clinical settings (Igaku-Shoin). - GENSHI AI will run a medical generative AI specialist exam and has opened pre-orders for the official textbook (Mainichi Shimbun, PR TIMES). - NTT, NTT Docomo Business and Igaku-Shoin are running a demonstration of a medical AI information platform linked with Ubie's generative AI (Iryo Tech News). - Japan's National Cybersecurity Office has asked for thorough measures against personal data leaks, including AI use and vulnerability assessments (Impress Watch). - Abroad: Thailand is promoting medical AI and precision medicine, targeting a health-related economy equivalent to 16 trillion yen by 2032 (newsclip.be).

Statements by key figures

No notable developments.