Diagnostics, clinical documentation and drug discovery — where AI meets medicine, with patient safety on the line.
217 articles
Researchers successfully utilized an OpenAI reasoning model to assist physicians in diagnosing complex, previously unsolved cases of rare genetic diseases. The study resulted in 18 new confirmed diagnoses, demonstrating the potential of reasoning models in clinical diagnostics.
OpenAI News · 93d ago
A UC Davis study reports an ALS patient using a brain implant to communicate independently for over 3,800 hours. The system achieved 99% accuracy and allowed the user to work full-time.
The Next Web (TNW) · 95d ago
NHS England is launching Microsoft 365 Copilot to over 505,000 staff members, marking the largest healthcare AI deployment globally. This follows a successful pilot program where 30,000 workers reported saving an average of 43 minutes per day on administrative tasks.
The Next Web (TNW) · 97d ago
NVIDIA introduces a new approach to evaluating clinical Automatic Speech Recognition (ASR) models using Agent Skills and Nemotron Speech. This solution addresses the difficulty of recognizing complex medical terminology, such as drug and procedure names, which often challenge off-the-shelf systems.
NVIDIA Developer Blog · 102d ago
MediaLink successfully implemented an AI chatbot for a Himeji emergency hospital, achieving 90% accuracy within two weeks. The project highlights the importance of a 'knowledge circulation' framework for effective AI deployment.
ITmedia AI+ (日本) · 104d ago
Traj-Evolve is a new multi-agent system designed to model longitudinal EHR data for lung cancer detection. It utilizes an Experience Pool for retrieval-augmented reasoning and multi-agent reinforcement learning to optimize inter-agent collaboration.
ArXiv AI · 108d ago
Researchers introduced a framework using LLM-guided evolutionary code search to optimize healthcare policy mechanisms. By simulating strategic provider responses, the system successfully mitigates up-coding and patient selection bias while maintaining financial efficiency.
ArXiv AI · 110d ago
EHRBench is a new, automated, and reliable benchmark designed to evaluate LLM performance in clinical decision-making using real-world electronic health records. It features nearly 1 million QA items across diagnosis, treatment, and prognosis tasks, verified through a knowledge-base pipeline.
ArXiv AI · 110d ago
China's new 'Prescription Drug Online Retail Compliance Guidelines' strictly mandates that prescription reviews must be conducted by licensed pharmacists, explicitly prohibiting AI from replacing human oversight. This move aims to curb the widespread 'instant prescription' and AI-driven prescription generation issues in the online pharmacy sector.
虎嗅 · 117d ago
Baichuan Intelligence has shifted its focus entirely to medical AI, launching the M4 medical model and the 'Bai Xiaoyi' agent to provide clinical decision support and patient health management.
36氪 · 119d ago
Corti has released Symphony for Speech-to-Text, a specialized AI model designed for clinical environments that significantly outperforms general-purpose models like OpenAI's Whisper in medical terminology. The model aims to provide a reliable data foundation for autonomous AI agents in healthcare workflows.
VentureBeat · 122d ago
An audit in Ontario revealed that 60% of AI-powered medical scribe systems frequently misidentify prescribed medications in patient notes. These findings highlight significant reliability concerns for AI tools deployed in high-stakes clinical environments.
The Register - AI/ML · 128d ago
The implementation of the '818 Order' marks a turning point for China's cell and gene therapy industry, ending a decade of regulatory gray areas. The industry is shifting toward a dual-track system of medical technology and pharmaceutical regulation.
虎嗅 · 128d ago
Corti launched a no-equity accelerator for healthcare and life sciences startups, granting access to its Symphony clinical-AI model stack. Symphony has outscored OpenAI on the HealthBench Professional benchmark.
The Next Web (TNW) · 129d ago
A study on MedSyn shows that interactive LLM support significantly improves diagnostic accuracy for emergency physicians. Residents experienced the most notable gains in diagnostic correctness when using the system as an iterative query tool.
ArXiv AI · 129d ago
New arXiv paper argues current AI benchmarks fail to measure reliability, safety, and clinical relevance in real-world healthcare workflows. Frontier models ace exams but score low (0.53-0.85) on documentation, decision support, and admin tasks.
ArXiv AI · 129d ago
Medicare introduces ACCESS, a new payment model designed specifically for AI agents in patient care. It creates the first governmental reimbursement mechanism for AI handling monitoring between visits, check-ins, housing referrals, and medication adherence.
TechCrunch AI · 129d ago
HypEHR is a compact Lorentzian model that embeds clinical codes, patient visits, and questions in hyperbolic space for efficient EHR question answering. It employs geometry-consistent cross-attention with type-specific pointer heads and is pretrained via next-visit prediction and ICD hierarchy regularization.
ArXiv AI · 148d ago
New York startup Coral has raised $12.5M to scale its AI platform that automates healthcare administrative tasks. The AI reads handwritten fax forms, processes prior authorizations, and completes patient intakes in under five minutes without requiring providers to change workflows.
The Next Web (TNW) · 152d ago
DeepER-Med introduces an agentic AI framework for evidence-based medical research with modules for planning, collaboration, and synthesis. It includes DeepER-MedQA, a dataset of 100 expert-level questions from real scenarios.
ArXiv AI · 152d ago
Researchers at Stanford developed a multi-agent AI system to automate patient chart summaries for Thoracic Tumor Boards, improving on manual workflows. The system was rigorously evaluated against physician gold standards using fact-based rubrics and LLM judges.
ArXiv AI · 156d ago
Google has discontinued its AI search feature 'What People Suggest', which provided crowdsourced health advice from amateurs worldwide. The move comes amid mounting scrutiny over AI delivering health tips.
The Guardian Technology · 187d ago
Eli Lilly partners Nvidia on $1B AI drug discovery lab. Anthropic launches Claude for Healthcare post-OpenAI's ChatGPT for Health.
虎嗅 · 191d ago
Amazon's Health AI assistant is now available to all US customers via its website and app. It no longer requires a One Medical subscription after the $3.9B acquisition.
The Next Web (TNW) · 192d ago
Sentinel, an autonomous AI agent, uses Model Context Protocol (MCP) for reliable triage of remote patient monitoring vitals, outperforming individual clinicians in emergency sensitivity (97.5% vs 60%). It achieves 95.8% emergency sensitivity against human consensus at $0.34 per triage.
ArXiv AI · 192d ago
Fudan University Affiliated Obstetrics and Gynecology Hospital and Alibaba Cloud launched the “Red House · Qiyuan” AI vertical large model for gynecology and obstetrics. The based “Xiao Hong” AI patient assistant passed Shanghai generative AI filing.
36氪 · 194d ago
AWS is launching Amazon Connect Health, a new AI agent platform tailored for healthcare. It helps with patient scheduling, documentation, and patient verification.
TechCrunch AI · 198d ago
AWS is launching Amazon Connect Health, an AI service for handling patient calls, clinical documentation, and billing codes. It directly challenges Microsoft's Nuance and competing AI scribe startups.
GeekWire · 198d ago
Amazon's cloud unit is launching AI tools for medical practices targeting patients and doctors. This initiative aims to expand AWS's cloud-computing business into workplace applications.
Bloomberg Technology · 198d ago
ChatGPT Health now integrates with Epic, allowing clinicians to import patient data into the product. OpenAI said the integration provides read-only access to health records.
TechCrunch AI · 18d ago