Oncology AI is moving beyond isolated predictions toward workflows that characterize disease, quantify response, personalize risk, and expose uncertainty. The defining challenge is becoming reliable clinical integration, not…

Respiratory AI is moving beyond one-off measurements: toward multi-night sleep assessment, multimodal cancer risk, longitudinal prediction, and systems-level maps that must still prove their value in prospective care.

PREDICT-GBM makes glioblastoma recurrence prediction more reproducible—and puts its current gains in perspective. At matched target volumes, the best-performing U-Net increased mean enhancing-recurrence coverage by 2.03 percentage points,…

AbdomenNet detected 11 acute abdominal conditions on noncontrast CT and improved radiologist performance in a retrospective reader study. Simulated worklist prioritization suggested faster reporting for emergent cases, alongside…

Emergency AI is expanding from isolated predictions into connected workflows spanning first contact, clinical action, quality surveillance, and model maintenance. The central challenge is turning benchmark gains into…

Integrating environmental, antibody, cytokine and gene-expression data revealed distinct signatures associated with paediatric atopic dermatitis (AD) and healthy status in South African children. These patterns suggest relationships to…

A single prompt-driven model approached task-specific nnUNet performance across diverse tumour segmentation tasks. SAT3D combines three-dimensional image analysis with voxel-wise confidence maps to guide iterative refinement, with encouraging…

Radiology & Imaging is moving beyond single-image detection toward longitudinal forecasting, guided acquisition, and foundation models designed for privacy, equity, and cross-site adaptation.

Recent coverage shows medical AI moving upstream into treatment selection, prognostication, and screening. The decisive question is shifting from whether models can detect signal to whether that signal…

Medical AI is getting better at extracting signal from ECGs, scans, biomarkers and clinical prose. This week's strongest papers sharpen a harder question: when does better prediction justify…