General-purpose large language model embeddings can encode longitudinal electronic health records with performance comparable to a specialized EHR foundation model. In this computational study, Qwen3-Embedding-8B matched CLMBR-T-Base on…

Recent Digital Health & Informatics coverage shows a field moving past plausible AI outputs toward workflow-tested tools, opportunistic biomarkers, secure imaging data, and less invasive sensing.

This week’s strongest medical AI papers were less about one-off detection and more about measurement infrastructure: sleep, CT, pathology, oncology, and body composition models that could reshape how…

Transcript INTRO Welcome to Medical AI Digest Podcast. Today, a paper from Nature biomedical engineering. MAIN Today we’re looking at “BoneCoT: multicentre validation of a whole-body skeleton foundation…

Transcript INTRO Welcome to Medical AI Digest Podcast. Today, a paper from The Lancet. Obstetrics, gynaecology, & women's health. MAIN The paper is titled “Measuring large language model…

Recent Endocrinology & Diabetes coverage points to a field shifting from static thresholds toward earlier, lower-burden risk detection across sensors, routine data, and molecular signatures.

Retinal fundus images may carry measurable signals of dementia risk. In this cohort study, a foundation model applied to routine eye photographs detected existing dementia and stratified future…

A voice-based AI assistant completed most pre-procedural preparation calls for cardiac catheterization while preserving nurse escalation for clinical issues.

Low-quality slit-lamp images can undermine automated keratitis diagnosis. Jiang, Song, Yao et al. report a two-stage AI framework designed to recover image quality and preserve diagnostic performance.

A large multi-omics analysis identifies SCG2 in tumor-associated macrophages as a key mediator of 5-FU resistance in colorectal cancer.