AI in Medicine News — 2026-10-23 (Fri) Daily News

Overview

Diagnosis moved most visibly: a Microsoft study reported an AI system outperforming physicians on diagnostic accuracy, and the FDA granted 510(k) clearance for the BoneEnhance machine learning module. Hospital AI showed documentation and operational gains alongside cost and reliability caveats, while drug discovery AI remained at the platform and discovery stage. On governance, South Korea and the WHO pledged equal access to medical AI benefits, and an FDA leader called for evaluating generative AI the way human physicians are evaluated.

Diagnosis and imaging

- A Microsoft study reported an AI system outperforming physicians on diagnostic accuracy; no accuracy figures or case counts are given in the headlines (Medical Economics, The Verge). - A radiology commentary argued the specialty's next step is facilitating data orchestration rather than fragmentation (Diagnostic Imaging). - Coverage asked whether AI can diagnose skin cancer better than a dermatologist and suggested the answer may surprise (Miami Herald). - Modern Healthcare's AI tracker reported that Baba secured $40M (Modern Healthcare).

Clinical decision support

- AI could bring clinical trials to rural America (Newsweek). - Stigmatizing patient details can push AI to deprioritize high-risk emergency cases, a bias finding in emergency-case prioritization (Medical Xpress). - South Korea's KT aims to replace over 50% of tertiary hospital medical data labeling with AI (BigGo Finance). - A new AI tool uses facial photos to predict cancer patients' survival risk; the evidence stage is not stated (Health and Me). - Burjeel Hospital partnered with South Korea's ARPI to advance AI-powered cardiac diagnostics (Gulf News). 【Generative AI and medical documentation】 - Using AI saves doctors time on documentation but won't fix physician burnout, per the analysis (The Conversation). - An updated guide to evaluating AI scribes for ambient documentation was published (Telehealth.org). - Hospital AI is delivering gains, but costs and setbacks expose implementation challenges (Korea Biomedical Review). - Three months in, HCTI's Purpose Health reported multiple wins as AI-led momentum builds (Yahoo Finance). - M42 launched its AI-powered Ossi platform at the Future Health Summit (WAM). 【Hospital operations and care delivery】 - Samsung Medical Center ranked 4th in a global AI smart hospital evaluation (The Chosun Ilbo). - GDHS 2026 will focus on implementing AI action (The Economic Times). - New AI heart technology in Batavia could cut hospital stays and invasive tests; the headline states this as a possibility, not a result (RochesterFirst). - 5G and AI are expected to make mobile hospitals a reality (The Korea Economic Daily). - Palantir is expanding AI deployments across defense, health, and construction (StreetInsider). 【Drug discovery and clinical trials】 No clinical-trial phase results appear in these headlines; the items are discovery-stage or platform and infrastructure work, with no preclinical, Phase I, or later data cited. - Scientists are already using AI agents, but most of what they learn goes unpublished (Drug Target Review). - Federated AI is being applied to drug discovery's data bottleneck (BioWorld News). - Reflector Bio emerged from stealth with an AI platform for drug discovery (citybiz). - A commentary explains the importance of ontologies to AI in drug discovery (Scientific Computing World). 【Medical device approval and regulation】 - FDA 510(k) clearance (clearance, item: BoneEnhance, the first machine learning layer on RIVANNA's Accuro XV platform) (PR Newswire, Unite.AI). - Drug Topics examined pharmacist roles and risks in guiding AI-enabled medical device use (Drug Topics). - Eko Health launched an AI-enabled stethoscope; regulatory status is not stated (MobiHealthNews). - DataMEDS AI entered into an amended and restated LOI to license and acquire IP covering goal-driven prescription drug telehealth, medical drones, and healthcare blockchain technologies; a letter of intent, not a completed transaction (The Des Moines Register).

Safety, errors and bias

- Stigmatizing patient details can push AI to deprioritize high-risk emergency cases (Medical Xpress). - A declaration with the WHO pledges equal access to medical AI benefits; Korean coverage identifies it as the Seoul Declaration (Korea.net). - The STEWARD project seeks to create oversight for AI in healthcare (The Stanford Daily). - Commentary argues it is time to invest in the "human stack" to foster trust in health AI (Healthcare IT News). - The Petrie-Flom Center issued a call for abstracts for its 2027 annual conference on generative AI in healthcare: law and regulation (Petrie-Flom Center). 【Research findings (major journals)】 - People are taking an unapproved weight-loss drug, and AI analysis found something surprising; the finding itself is not specified in the headline (Science Daily). - Microsoft and Opella are partnering on an agentic AI content verification platform (Fierce Healthcare). - Coverage examined what patients think of AI in patient portal messages (TechTarget).

Developments in Japan

- LSTC and four other parties will jointly develop three systems to speed up medical AI using 2nm-generation semiconductors; the project was adopted by Japan's NEDO (NOVAIST). - Anno of Japan's Team Mirai said damage involving stoppages of electricity, water, and medical services should be assumed, and will make an urgent proposal to Japan's Digital Agency, after a cyberattack exploiting AI (ITmedia). - Singapore's Economic Development Board published Japanese-language coverage of policies and research infrastructure supporting the practical application of AI healthcare, precision medicine, and healthy longevity technologies (Singapore Economic Development Board). - Entertainment coverage marked the premiere of the medical drama Analog Nurse, in which Nao Matsushita plays a top freelance nurse who sticks to an analog approach at a large hospital that relies on AI (Yahoo! News). 【Statements by prominent figures】 - Mustafa Suleyman of Microsoft said its AI outperforms doctors on hard cases as he pursues what he calls "medical superintelligence" (TechCrunch). - An FDA leader with a radiology background said generative AI should be evaluated the way human physicians are (Radiology Business).