Every story in this issue with a short summary, so you can decide what to read before leaving the site. Summaries are written by AI (Claude) only from each publisher's own description and checked by Jev for anything not in that description. They are not full translations. Follow "Read the original" for the full story.
A narrative review in Discover Artificial Intelligence maps evidence gaps for hybrid CNN-ViT architectures in mammography. It highlights the disparity between retrospective benchmark performance and other evidence.
Read the original →A new deep learning framework called Ex-CMRNet uses hybrid encoders and LIME explainability. It detects myocardial inflammation in cardiac MRI images with high accuracy.
Read the original →A guide to AI clinical decision support systems covers FDA Section 520(o)(1)(E) non-device criteria and ONC HTI-1 45 CFR 170.315(b)(11) DSI. It also addresses explainable AI and bias.
Read the original →Researchers developed an interpretable machine learning model and a web-based tool. They forecast postoperative in-hospital mortality in patients with acute aortic dissection.
Read the original →A technical guide to Infermedica and Kahun AI covers Bayesian diagnostic networks, 5-tier triage APIs, literature-mapped knowledge graphs and HL7 FHIR integration.
Read the original →A European survey of 874 patient community respondents found widespread recognition of the benefits of healthcare AI. Respondents also expressed significant concerns about empathy and diagnostic safety.
Read the original →A technical guide to Twofold AI Medical Scribe (Twofold Health) covers SOAP, DAP, BIRP and ADIME templates, ambient and dictation modes, HIPAA BAA, and EHR transfer.
Read the original →The item covers AI governance in healthcare, including the 6 WHO ethical principles (2021) and the 2024 WHO Large Multi-Modal Model (LMM) guidance. It also addresses validation by hospital AI committees.
Read the original →Messages released under freedom of information rules show Palantir telling former health minister Zubir Ahmed that adding GP records was unarguable. This came despite a 2023 pledge against it.
Read the original →3billion has signed a Master Co-Development Agreement with DeepLife. The deal combines genomic data from more than 120,000 individuals with AI virtual-cell technology for therapeutic target discovery, validation, and drug candidate development.
Read the original →PSC Biotech announced its appointment as the sole Sales Agent for Clever Culture Systems’ APAS® Independence across the pharmaceutical sector.
Read the original →Swiss venture firm Redalpine has closed its oversubscribed eighth early-stage fund at $250 million. The first investments are Enact, Helical and Arlequin AI.
Read the original →Boltz PBC has partnered with Excelra to power its AI drug discovery models with GOSTAR™ data. The data includes curated GOSTAR™ measurements of kinetic properties.
Read the original →Artificial intelligence is reshaping bioprospecting, a process that has historically relied on labor-intensive extraction and screening of natural compounds.
Read the original →The guide covers FDA artificial intelligence regulation, including the list of AI/ML-enabled medical devices and the 510(k), De Novo and PMA pathways. It also addresses GMLP principles and Non-Device CDS.
Read the original →Global investment in AI drug development has doubled to $8.4 billion in two years. No AI-designed drugs have FDA approval, and only two have reached phase 3.
Read the original →The item compares AMA Policy H-480.940 on augmented intelligence with autonomous clinical AI. The comparison covers FDA 520(o)(1)(E) clinical decision support criteria, liability, and override controls.
Read the original →Founder Akash Kapoor left medical school in his final year to help AI developers teach their models better clinical judgment.
Read the original →Clalit Jerusalem District chief Dr. Yael Reichenberg says AI can help doctors predict and prevent disease. Google.org’s decision to work with the Israeli health provider signals growing international recognition of its medical AI capabilities, according to the description.
Read the original →The item explains how the Coalition for Health AI (CHAI) Blueprint, CHAI Applied Model Cards, Assurance Labs, and the Health AI Partnership (HAIP) govern clinical AI.
Read the original →Meituan released the weights for its LongCat-2.5 artificial intelligence model on Nov. 9 under the MIT License.
Read the original →A comparative study evaluated five popular artificial intelligence models answering patient questions about irritable bowel syndrome. It found notable differences in information quality.
Read the original →He has urged his followers not to engage with people impersonating him online or send them money.
Read the original →AI is changing drug discovery and medical practice, with results such as early cancer detection and shorter drug development times through analysis of large volumes of data. However, creating "new knowledge" that goes beyond existing findings is considered difficult. Some opinions point to limits, saying AI cannot replace humans.
Read the original →UiPath (PATH) has become a launch partner for Google Cloud's Gemini for Healthcare. The integration works through UiPath's Model Context Protocol Server.
Read the original →SORACOM will begin offering its IoT and Physical AI System Security Assessment and Hardening Service on November 10. The service targets IoT, physical AI and other network-connected systems.
Read the original →