A multiplex plasma proteomics panel combined with supervised machine learning showed promise for distinguishing major dementia etiologies in a cohort study.

AI-assisted dermoscopy improved dermatology residents’ diagnostic accuracy for challenging facial pigmented lesions, particularly lentigo maligna.

A new Radiology: Artificial Intelligence study shows that chest radiograph AI can exploit acquisition exposure parameters as hidden shortcut features, with clinically meaningful performance bias.

In the LUNA25 challenge, a top-performing AI system outperformed the average radiologist in estimating malignancy risk for indeterminate 5-15 mm lung nodules on screening CT.

A large linked-record study has produced a validated tool to estimate serious muscle disorder risk in people eligible for statins in England.

A data-efficient annotation loop using nnU-Net substantially reduced expert time and estimated costs for medical image segmentation across diverse CT and MRI tasks.

Near-real-time AI abstraction paired with physician feedback improved performance on a national sepsis quality measure in a randomized trial.

This study tested whether depression progression can be divided into actionable biological states before full disease onset. In mice, an AI-guided multimodal framework identified a reversible pre-disease window…

A large unsupervised EHR analysis identified clinically distinct patient groups and showed that earlier onset of common phenotypes tracks with mortality risk and premature aging.

AI-driven cuffless blood pressure monitoring is advancing rapidly, but this systematic review argues that the evidence base is not yet strong enough for trustworthy clinical use.