A new smartphone system accurately measures human movement at laboratory-grade precision, making objective gait analysis accessible in routine clinical practice.

A new study evaluates ChatGPT-4o's ability to diagnose complex polyneuropathies, finding it can assist non-specialists and suggest appropriate confirmatory tests, but cannot yet replace specialist neurologists.

ProViCNet, a deep learning model using patch-level contrastive learning, outperformed expert radiologists in detecting clinically significant prostate cancer on MRI. When combined with PSA screening, it increased diagnostic…

EyeRAG, a knowledge graph-augmented AI system, cuts hallucinations to 3.3%—a 90% improvement over standard models—for safe patient counseling in ophthalmology.

A machine learning model trained on routine blood indices accurately predicts colorectal cancer survival and outperforms traditional TNM staging, offering an accessible prognostic tool for clinical practice.

A new vision-language model trained via curriculum learning matches how radiologists develop expertise, achieving state-of-the-art performance in complex diagnostic tasks without task-specific fine-tuning.

Deep brain stimulation reduces Parkinson's motor symptoms in most patients, yet quality-of-life outcomes vary dramatically. A new study identifies three key predictors: younger age, emotional burden, and precise…

Researchers validated an AI-assisted mobile application that enhanced generalist healthcare professionals' ability to identify skin cancer risk, reducing unnecessary specialist referrals by 30% while maintaining diagnostic accuracy in…

SlAction, an infrared video system, detects sleep apnea without sensors or cloud data, processing entirely on-device.

A cross-national study of over 3,600 participants found that physicians accepting standard-care AI recommendations face no increased malpractice liability in either jury-based or expert-based legal systems.