AI in Medicine News — 2026-10-27 (Tue)

Overview

Activity centered on diagnostic imaging and pathology, hospital operations and governance, and regulation of AI-enabled medical devices. Evidence on LLM-based diagnosis stayed at the research stage, with reports that Microsoft's AI has not yet faced real patients (www.techtimes.co.uk, TechRepublic). Safety and oversight themes ran through several items, including a data-handling incident and a first WHO healthcare AI declaration (Straight Arrow, Healthcare IT News).

Diagnostic imaging

- A Science Daily headline says AI detects "invisible" signs of pancreatic cancer years before diagnosis. The headline gives no detection rate or evidence stage (Science Daily). - An AI model uses routine pathology slides to predict breast cancer recurrence risk. The headline gives no performance figures or study stage (Clinical Lab Products). - Agendia unveiled clinical validation of its AI-driven digital pathology approach. No metrics appear in the headline (BioSpace). - A Korean AI for Alzheimer's diagnosis and an artery analysis tool received 510(k) clearance, according to a briefs roundup (MobiHealthNews). - Diagnostic Imaging published a roundup titled "Advances in AI — October 2026" (Diagnostic Imaging).

Clinical decision support

- A machine learning model predicts 28-day mortality in sepsis patients with diabetic retinopathy. The headline does not state validation status (Bioengineer.org). - NVIDIA invested in Doctolib to speed healthcare AI (Konsulteer). - A headline cites over 1,000 FDA-cleared AI devices and asks where they actually work in hospitals (Intelligent Living). - Medical Buyer reports that India's healthcare is reaping AI boom dividends (Medical Buyer). - A Tech Policy Press piece warns that AI partnerships in Africa without an exit plan risk becoming the new tied aid (Tech Policy Press). 【Generative AI and clinical documentation】 - eClinicalWorks offers new AI-enabled EHR tools (Healthcare IT News). - Two articles cover AI scribes for patient communication and AI for clinical workflows among family physicians (The Cardiology Advisor). - Mulk Technology launched an AI-powered digital hospital platform for India (Express Healthcare). 【Hospital operations and care delivery】 - Michigan Medicine's new AI chief joins what the headline calls a fast-growing club of hospital executives (Crain's Detroit Business). - Cleveland Clinic's AI task force developed a unified strategy to integrate technology innovations (Cleveland Clinic). - A hospital employee uploaded patient data to an unapproved AI tool, and the headline asks where the data is now (Straight Arrow). - Sentara hospitals are using AI to help monitor patients and determine discharge timelines (Virginia Mercury). - HCI Innovation Group argues that governing healthcare AI means putting patient safety first (HCI Innovation Group). 【Drug discovery and clinical trials】 - Cleveland Clinic and Dyania's work on AI in clinical trial matching earned a 2026 Gartner Eye on Innovation Award. The headline does not give a trial phase (Cleveland Clinic Newsroom). - MitoCareX Bio completed the first AI drug-discovery screening stage. This is an early discovery stage, not clinical (StreetInsider). - Florence is automating the connection between study sites and sponsors, with AI tools already live (Chinook Observer). - The Recursive asks, in a headline, who pays when a drug must answer 5 questions and AI made only one cheap (The Recursive). - PA Future says AI is reshaping healthcare economics from drug discovery to patient care (PA Future). 【Medical device approval and regulation】 - Guidance/policy request: ATA Action urged FDA to build a risk-based framework for AI-enabled medical devices (HME News). - Clearance: 510(k) clearance for a Korean Alzheimer's diagnosis AI and an artery analysis tool; product names are not given in the headline (MobiHealthNews). - Litigation: Heartflow took to the courts over AI CCTA technology (Cardiac Wire). - Istituto Impariamo says FDA has cleared over 1,500 medical AI devices and stresses that the difference between clearance and real clinical proof is staggering. Another outlet cites over 1,000 cleared devices, so the counts differ by source (Istituto Impariamo, Intelligent Living). - Drug Topics covers how AI-enabled devices are changing pharmacy (Drug Topics).

Safety, errors and bias

- Petrie-Flom Center argues that clinical AI should earn its autonomy the way doctors do (Petrie-Flom Center). - WHO and Korea signed the first healthcare AI declaration (Healthcare IT News). - ORF discusses what AI prescribing means for public health (orfonline.org). - Microsoft says its AI diagnoses better than doctors, but it has not faced real patients yet (www.techtimes.co.uk). - Vietnam.vn states that lesion-detecting AI still needs verification by doctors (Vietnam.vn). - The unapproved AI tool data upload noted above is a data-governance concern (Straight Arrow).

Research findings

- LLMs recommend far more care than doctors. The headline gives no study design (Digital Health Wire). - AI chatbots may worsen symptoms in people with psychosis (Medical Xpress). - Microsoft's AI diagnoses 83% of complex cases, with the headline asking whether it can work in hospitals. This is a research result, not hospital use (TechRepublic). - adsilico raised £10M to test medical devices on AI-generated virtual patients (Dealroom). - Pharmaceutical Commerce covers how autonomous AI is reshaping biopharma operations (Pharmaceutical Commerce).

Developments in Japan

- The headline gives no evaluation details (Medical Tech News). - In Microsoft's stepwise diagnosis study, AI had an 83% correct diagnosis rate on 304 difficult cases versus 36% for physicians (NOVAIST). - Nikkei Tech Foresight reports on medical AI using 2-nanometer semiconductors, involving LSTC (including Rapidus participation), with chips from IBM (Nikkei). - A Japan-origin startup is working to export Japanese healthcare as a "medical AI advanced nation" (DG Lab Haus). - Impress released "AI Gyomu Katsuyo Navi", a service for small and mid-sized companies to assess where to begin with generative AI. It is not specific to healthcare (Impress).

Statements by key figures

No notable developments.