A fine-tuned deep neural network staged sleep in Parkinson’s disease and isolated REM sleep behavior disorder with agreement comparable to human interrater performance. Its confidence scores also provided…

In the CADLY2 randomised trial, computer-aided detection did not significantly increase adenoma detection during high-quality colonoscopic surveillance of people with Lynch syndrome. Computer-aided diagnosis also remained less reliable…

SmartAlert reduced repetitive inpatient complete blood count testing by identifying results likely to remain clinically stable and intervening directly within the ordering workflow. In a randomized pilot, the…

General-purpose large language models became a substantial source of informal mental health support in 2024. An estimated 14–18 million U.S. adults may have used them for this purpose,…

FAST-NSTEMI combined high-sensitivity cardiac troponin T with readily available clinical variables to identify or exclude NSTEMI. Its main advantage was greater triage efficacy, particularly from a single blood…

Maternal metabolism may reveal more than the risk of an early delivery: it may also distinguish how that delivery arises. In this cohort of African American women, spontaneous…

Pathology AI is moving beyond classification toward tissue atlases, latent biomarkers, and molecular discovery. The decisive question is whether these signals remain biological—and clinically useful—across centers, cohorts, and…

In a six-week randomized trial, a purpose-built generative AI app strengthened some aspects of student well-being and helped preserve others. The findings position Flourish as a low-intensity, preventive…

Can surgical-video AI learn across hospitals without moving sensitive recordings between them? In a multinational appendectomy cohort, weakly supervised Swarm Learning achieved external performance close to centralized training…

The TopCoW Challenge shows that automated Circle of Willis analysis can move beyond identifying vessel voxels to preserving the vascular connections that define clinically relevant anatomy. Leading algorithms…