AI in Medicine News — 2026-10-16 (Fri) Morning News

Overview

Diagnostics and device clearance, drug discovery and hospital operations produced the most headlines. Two FDA-related items concern breast cancer recurrence AI tools (ABC News, Unite.AI), and AstraZeneca announced an AI drug discovery collaboration with insitro (Reuters, Fierce Biotech, AstraZeneca). Governance frameworks for agentic AI and a UK shift to lifecycle monitoring also appeared (Yahoo, Burges Salmon).

Diagnostics and imaging

- A new FDA-cleared AI tool could help predict whether breast cancer will return. The headline gives no performance figures (ABC News). - Eyonis will present clinical evidence for AI-powered lung cancer screening at CHEST 2026. The headline does not state the results (GlobeNewswire). - NeoGenomics expanded its Indica Labs partnership to scale AI-powered digital pathology (Business Wire). - An AI checklist helps radiologists read tricky heel surgery X-rays faster and more accurately. No figures are given in the headline (Bioengineer.org). - Agilent will acquire Proscia, a Philadelphia company, in a deal worth up to $270 million, expanding AI-enabled digital pathology (BioBuzz).

Clinical decision support

Safety, errors and bias

- Healthcare workers are using AI to create more time with patients, according to research. The headline does not state the study design (Spectrum News). - One piece discusses AI in healthcare and the evidence that should guide adoption (Innovation News Network). - Endpoints News reports on San Francisco's pulse on health AI (Endpoints News). 【Research results (major journals)】 - An AI-powered mental health app demonstrated improved social well-being among users (2 Minute Medicine). - Penn engineers used AI and Reddit to identify potential side effects of an unapproved new GLP-1. This is a signal-detection study, not a clinical trial result (EurekAlert!). - The Lancet Commission on 17 Global Threats found that malicious AI and nuclear famine carry low-probability extinction risk this century (Nature World News). - Scientists prompted ChatGPT to mimic fear and sadness. The headline is truncated, so the finding is unclear (The Economic Times). - Patients' own alert labels helped an AI seizure monitor cut false alarms by about 80% in simulation. The evidence stage is simulation, not clinical use (inkl).

Developments in Japan

- Nikkei Medical asks whether risk management is adequate as generative AI spreads rapidly (Nikkei Medical). - Path Imaging is automating skilled techniques that support pathological diagnosis, with new technology for the AI medical era (Nikkei). - RetinaMind, a 17-year-old's AI, explores autism and ADHD from the retina. The headline gives no performance figures (Pasquale Pillitteri). - A Vietnamese associate professor won the top prize at an ASEAN-US science contest for AI research in healthcare. This is not a Japanese development (Vietnam.vn). - Fortience Consulting began offering an AI-native diagnostic service for manufacturers. This is not a medical item (Infoseek).

Statements by key figures

- DeepMind's Hassabis called AI agent hacks "Early Warning Shots" (Technology Org).