AI in Medicine News — 2026-10-22 (Thu) Daily News

Overview

Movement appeared across all five areas in the past 48 hours: diagnosis and imaging, clinical decision support, hospital operations, drug discovery and regulation. On the regulatory front, the FDA cleared two AI-enabled coronary flow software products (FFR-IQ and Elucid's FFR-CT), its discussion paper on generative AI-enabled medical devices drew public comment, and the AMA called for risk-based oversight, while US states are advancing AI regulation and South Korea and the WHO issued their first joint declaration on healthcare AI (Diagnostic Imaging, Cardiovascular Business, Bipartisan Policy Center, DOTmed; National Conference of State Legislatures; Korea Biomedical Review, MLex). In care delivery and drug discovery, Epic and Doximity pushed AI integrations for clinicians, health systems announced enterprise virtual care and pediatric AI partnerships, and Astellas committed $400M to a computational drug discovery deal with AI Proteins (Fierce Healthcare, Becker's Hospital Review, Hospital News, Fierce Biotech).

Diagnosis and imaging

Google Research reported work on AI-assisted ultrasound to support frontline maternal care (Google Research). The FDA cleared AI-enabled FFR-IQ software that estimates fractional flow reserve from CCTA; what is measured is the fractional flow reserve estimated from CT angiography images, with no performance figures given in the headline (Diagnostic Imaging). A supercomputer was used to help design more reliable AI for breast cancer diagnosis — a development-stage effort, with no accuracy level stated (News-Medical). One report says AI improves detection but may reshape clinical reasoning (Bioengineer.org). The British Institute of Radiology debuted its AI Academy, an education initiative rather than a diagnostic tool (Imaging Technology News).

Clinical decision support

Doximity added UpToDate to its Ask tool in a push to become the "go-to AI platform" for doctors (Fierce Healthcare). A study warns AI could push high-risk ER patients down the queue (Newsweek). California lawmakers are trying to set some limits as AI systems act as therapists and give out advice (News From The States). Commentary argues primary care AI must prove it improves the next clinical decision (Innovation News Network). Wolters Kluwer set out four dimensions of trust in clinical AI (Wolters Kluwer). 【Generative AI and medical documentation】 Epic launched Connectors to link its AI tools with third-party apps (Fierce Healthcare). M42 launched Ossi, an AI platform aimed at transforming clinician–patient connections (Abu Dhabi Media Office). 【Hospital operations and care delivery】 St. Luke's Health System selected hellocare.ai for an enterprise AI-assisted virtual care deployment in southern Idaho (Becker's Hospital Review). Ob Hospitalist Group was named a Newsweek AI Impact Health Award winner for Best Health System AI Transformation (PR Newswire). Three global hospital leaders say leadership must change, too, as AI changes hospitals (Korea Biomedical Review). SickKids and Microsoft announced a five-year partnership to advance AI-powered pediatric care (Hospital News). Alcidion launched an AI-enabled platform to overhaul hospital administration (Health Tech World). 【Drug discovery and clinical trials】 Stage: early research — AI designed functional enzyme components that could help expand drug discovery; no clinical testing is indicated (Phys.org). Astellas linked up with AI Proteins in a $400M computational drug discovery deal; no clinical trial stage is given for any program (Fierce Biotech). Legal commentary addresses AI-generated evidence in drug discovery, covering reliability, legal risk and FDA compliance (Crowell & Moring LLP). KraneShares argues China may control a key bottleneck in AI drug discovery (KraneShares). Healthcare and life sciences leaders are building on NVIDIA technology to advance drug discovery and expand patient care (NVIDIA Newsroom). 【Medical device approvals and regulation】 Clearance: the FDA cleared FFR-IQ, AI-enabled software for estimating fractional flow reserve from CCTA (Diagnostic Imaging). Clearance: the FDA cleared Elucid's new AI-enabled FFR-CT software (Cardiovascular Business). Draft guidance: IAM Media reports that AI-enabled medical devices require an integrated FDA and IP strategy following draft guidance on pre-market submissions; this is draft-stage guidance, not a final rule (IAM Media). Discussion paper: the FDA published a discussion paper on generative AI-enabled medical devices, on which the Bipartisan Policy Center submitted a comment (Bipartisan Policy Center). Policy call, not an approval or guidance: the AMA called for risk-based FDA oversight of generative AI medical devices (DOTmed).

Safety, errors and bias

Utah has approved 3 health AI pilot apps to run in a regulatory sandbox (BankInfoSecurity). A Chosun Ilbo piece argues public medical AI infrastructure is essential to prevent regional gaps (Chosun Ilbo). US states are leading the way on AI regulation while Congress deliberates (National Conference of State Legislatures). South Korea and the WHO issued their 1st joint declaration on healthcare AI and called for equitable access (Korea Biomedical Review). The WHO and South Korea are to cooperate on responsible, equitable use of healthcare AI (MLex). 【Research results (major journals)】 An AI model demonstrated the potential to improve the efficiency of screening for structural heart disease using electrocardiograms; this is a research demonstration of screening efficiency, with no accuracy figures given (2 Minute Medicine). An advanced AI tool estimates biological age and predicts cancer survival from face photos; this is a research-stage model, and no performance level is stated in the headline (News-Medical).

Developments in Japan

Nikkei Medical examined how far the world recognizes AI-based diagnosis and diagnostic support (Nikkei Medical). Protege began independent healthcare AI evaluation, and its Disease Odyssey tool revealed diagnostic challenges (Unite.AI). Marso launched a security support service for medical institutions that uses AI to diagnose website vulnerabilities (Medical Tech News). LSTC and Hokkaido University aim to develop medical AI using 2nm semiconductors and bring it into practical use (PlusWeb3). The TRIAD healthcare AI framework shifts the focus of evaluation from models to teams (remio). Also relevant to Japan, Astellas linked up with AI Proteins in a $400M computational drug discovery deal (Fierce Biotech).

Statements by key figures

A digital health leader at the FDA's Center for Devices and Radiological Health (CDRH) said generative AI policy is a White House priority (MedTech Dive). Three global hospital leaders said leadership must change, too, as AI changes hospitals; the headline does not name them or their titles (Korea Biomedical Review).