AI in Medicine News — 2026-10-12 (Mon) Morning News

Overview

Reports centered on diagnostics (ECG and breast ultrasound AI), clinical safety and bedside performance, and AI scribe governance. Several items concern regulation and oversight, including state lawmakers and a Lancet assessment of AI as a catastrophic threat. Drug discovery, clinical trial operations and hospital operations also featured, but the headlines give no accuracy figures or trial phases.

Diagnostics and imaging

- A deep learning system reads breast ultrasound videos "better than doctors," according to the headline. The headline gives no figures, and the study's stage of evidence is not stated (Bioengineer.org). - AI reads children's ECG reports to flag hidden heart disease. No accuracy figures or validation stage appear in the headline (Bioengineer.org). - A systematic review covers AI-enhanced electrocardiography for detecting hypertrophic cardiomyopathy in athletes, addressing diagnostic accuracy and external validity. The headline gives no results (Cureus). - A study of Google's AMIE is described as showing how AI could support doctors. The headline gives no performance level (WeRSM). - AnX Robotica highlighted documented ProScan AI workflow performance from submission to FDA. Regulatory status is covered in section 7 (PR Newswire).

Clinical decision support

- A roundup item reports that 49% of patients have administered a self-test and 43% have used an AI chatbot instead of seeing a doctor because of cost concerns (Healthcare IT Today). - Sutter Health describes how AI helps modernize mammography screening. The headline gives no performance data (Sutter Health). - AI maps ICU patients' physiological trajectories to predict deadly decline. The headline gives no accuracy figures or validation stage (Bioengineer.org). - Electronic health records are evolving in the AI era (Healthcare Digital). - Art Papier, M.D., warned dermatologists not to assume AI is always right, at Fall Clinical Derm 2026 (Managed Healthcare Executive). 【Generative AI and clinical documentation】 - A piece examines how regulators actually govern AI scribes in healthcare, using the framing "The Drafting Fiction" (Bioengineer.org). - Google's AMIE trial and Utah's AI healthcare pilots were covered together in a newsletter item (The Neuron). 【Hospital operations and care delivery】 - A patient reported paying a hospital bill, after which the hospital's AI demanded $27,000 more, in a story headlined "An 'Epic' failure" (Washington Examiner). - Medical AI is gaining ground in Chinese hospitals (Borneo Bulletin). - A report says Big Tech set a $715B-$740B 2026 spending range as healthcare AI expands (TOKENPOST). - An article argues that healthcare and the environment can benefit from "Green" AI (healthcare-in-europe.com). 【Drug discovery and clinical trials】 - GSK has a new AI drug discovery deal. The headline gives no program, target or development stage (Yahoo Finance). - AI-driven automation reduces EDC query volume as trial complexity grows (The Clinical Trial Vanguard). - Three converging waves, not just AI, are transforming clinical data (The Clinical Trial Vanguard). - Advarra funds a PRIM&R AI ethics curriculum for research oversight (The Clinical Trial Vanguard). 【Medical device approvals and regulation】 - ProScan AI (AnX Robotica): the headline refers to workflow performance "from Submission to FDA." It does not state whether this is a clearance, approval or designation, so the status is a submission-related workflow only (PR Newswire). - No other approvals, designations or guidance are named in the headlines.

Safety, errors and bias

- A study finds that high test scores do not translate to bedside performance, described as the hidden flaw in medical AI (Bioengineer.org). - A Mount Sinai study finds that safety prompts can help AI models make safer clinical choices. The headline gives no figures (Mount Sinai). - State lawmakers and business leaders are weighing regulation of AI in healthcare (Wisconsin State Journal). - The 11th Annual Malaysia OpenGov CXO Leadership Forum 2026 includes an AI in healthcare session. No content is given (OpenGov Asia). 【Research results (major journals)】 - A Lancet assessment says nuclear war and malicious use of AI threaten the end of humanity by 2100 (The Guardian). - The Lancet report also names AI-enabled bioweapons alongside nuclear war as threats that could wipe out humanity by 2100 (Hindustan Times). - AI is among 17 potentially "catastrophic threats" that researchers say require targeted action (CNN). - The report warns that AI and nuclear war threaten human extinction by 2100 (Sky News, Ukrainian National News). - These are risk assessments, not clinical studies of AI in medicine.

Developments in Japan

- The only medically relevant item is a Japanese-language report on US policy: CMS's health tech Slack offers AI companies a shortcut to Medicare policy (remio). - No notable developments in Japan's medical sector. The other Japanese-language items concern horse racing, fragrance, office equipment and AI security, and are unrelated to medicine.

Statements by key figures

- Art Papier, M.D. (title not given in the headline) warned dermatologists not to assume AI is always right (Managed Healthcare Executive). - No statements from tracked figures, since no list was provided.