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

Overview

The headlines center on diagnostic AI for neurology and oncology, safety of clinical AI, and AI in drug discovery and clinical trials. The most substantial items are the Mount Sinai safety-prompt result and a study finding weak enforcement of medical AI policies. None of the headlines states an approval or regulatory designation.

Diagnostics and imaging

- Lunit is looking to bring pathology AI into drug development through a US lab deal (Korea Biomedical Review). - An AI system that reads brain scans and medical records is reported to boost Alzheimer's diagnosis (Bioengineer.org). - AI helps neurologists distinguish epilepsy from functional seizures (News-Medical). - A Korean AI model spotted a new liver cancer biomarker (MobiHealthNews). - Google's AMIE chatbot interviewed 98 Boston primary care patients under doctor supervision without a single safety stop (Medical Daily). - The headlines give no accuracy or detection figures, and none states the stage of evidence.

Clinical decision support

- Mount Sinai reports that an AI safety prompt cuts harm 39% (shattered.io). The headline does not say which study design or setting this comes from. - RadNet's AI tool could help nearly twenty percent of patients avoid prostate biopsies (The Indian Practitioner). The wording is conditional, so this is a potential effect rather than an established result. - A risk-calibrated transformer is reported to make Alzheimer's staging safer by learning to say "I don't know" (Bioengineer.org). The headline gives no performance figures. - An op-ed argues that hospitals are scaling AI and that the workflow needs its own safety test (Eurasia Review). 【Generative AI and medical documentation】 - Healthleap raised $38M for hospital AI screening (streamlinefeed.co.ke). The headline describes screening, not documentation or clinical records. 【Hospital operations and care delivery】 - South Korea released an AI-powered care roadmap (Healthcare IT News). - Taiwan's health-tech sector, spanning hospital AI to home monitoring, is testing its US market fit in Dallas (GeneOnline). - A profile describes Mohammad Noshad as building AI that helps protect patients (The Good Men Project). The headline gives no detail on the product or its evidence. 【Drug discovery and clinical trials】 - AI adoption in clinical trials is hindered by validation and governance gaps (The Clinical Trial Vanguard). - GenScript and NVIDIA launched a global PitchFest to accelerate AI-driven drug discovery startups (BioPharma APAC). - Danaher is to launch an AI-powered autonomous lab (Scientist Live). - None of these headlines states a development stage such as preclinical or Phase I. 【Device approvals and regulation】 No notable developments.

Safety, errors and bias

- A study finds that global medical AI policies score well on vision but falter on enforcement (Bioengineer.org). - The AMIE, Mount Sinai and risk-calibrated Alzheimer's staging items above also bear on safety (Medical Daily, shattered.io, Bioengineer.org). 【Research results (major journals)】 - A Lancet analysis lists nuclear war and malicious AI among 17 health threats posing extinction risk by 2100 (Yahoo, Banglanews24, DongA Science, bdnews24.com, Unboxholics). The headlines do not describe the methodology.

Japan

No notable developments.

Statements by key figures

- Shou Chew, TikTok CEO, expects major scientific breakthroughs from AI within three years (CNBC, The Next Web). The headlines do not tie this to medicine specifically. - Dr. Simeon George of SR One discussed AI, China and biotech investment on Business Of Biotech (BigGo Finance). The headline gives no further detail.