A brief resting-state EEG may help identify amyloid positivity in cognitively normal older adults, according to an exploratory study of 44 participants. The most striking reported accuracies depended…
A plasma autoantibody signature distinguished esophageal squamous cell carcinoma from normal controls across validation cohorts, including a small prediagnostic group. Its sensitivity was moderate, however, leaving endoscopy essential…
QoQ-Med3, a multimodal clinical model trained on public datasets, improved benchmark performance and transferred to electronic health record tasks. Persistent errors and uneven performance across some demographic groups…
Blood-based measures of aging showed different patterns in patients with endometriosis and those with endometriosis-associated ovarian cancer. Age-related methylation stood out in the primary comparison, but the results…
Genomic surveillance of multidrug-resistant Salmonella across China’s food chain has identified a large pool of high-risk resistance traits, most associated with mobile genetic elements. Projections suggest many of…
National cancer control plans often describe implementation and health services more fully than financing or measurable outcomes. An AI-assisted analysis of plans from 50 countries found a median…
A medical AI update can be acceptable for deployment while a separate alarm signals potential risk to the model already in use. AEGIS demonstrates how those decisions could…
AI can help clinicians detect intracranial haemorrhage on head CT, especially when its predictions include visual explanations. But accepting incorrect advice is only part of the problem: rejecting…
Combining histopathology foundation models improved diagnostic agreement for esophageal cancer detection, but did not overcome limited sensitivity for cancer and dysplasia. The findings support further study of these…
In a randomized diabetes prevention trial, participants who started AI-led coaching did so sooner than those assigned to human coaching. But after 12 months, survey respondents rated human…