MedAI Digest
A Photonic Network Laser Shows Promise for Learning from Scarce Medical Images
A retinally inspired photonic network used competing laser modes to extract image features, reaching 77.7% breast histopathology classification accuracy with just 10 training images. The laboratory proof of…
Digital Health & Informatics: From Model Performance to Clinical Fit
The field is shifting from benchmark gains to clinical fit: safe generation, deployment-aware prediction, closed-loop wearables, and context-rich forecasting now hinge on workflow, transportability, and actionable performance.
Faster Rheumatology Access, but Triage Still Misses One in Three Inflammatory Cases
A structured triage pathway helped patients with inflammatory rheumatic and musculoskeletal diseases reach comprehensive assessment sooner. However, the brief rheumatologist assessment failed to suspect nearly one-third of ultimately…
AI Detects Active and Structural Sacroiliac MRI Lesions Across Three Phase III Trials
A fully automated deep-learning pipeline detected active and structural sacroiliac-joint MRI lesions with performance generally comparable to expert interreader agreement. Crucially, models trained on one phase III trial…
Surgery: AI Moves From Risk Forecasts to Operative Decision Support
Recent Surgery coverage shows AI moving across the care pathway: forecasting risk before intervention, interpreting operative video, standardizing skills assessment, and exposing access barriers that algorithms cannot afford…
An Exosomal Signal of Hidden Liver Metastasis in Pancreatic Cancer
A plasma-based signature of seven exosomal microRNAs identified pancreatic cancers likely to recur in the liver within six months of resection. Its performance remained strong in two external…
Pulmonary Nodule AI Cut Chest CT Reporting Time by 15% in Routine Practice
Commercial pulmonary nodule AI was associated with a meaningful reduction in chest CT reporting time in routine clinical practice. Yet the benefit was far from uniform: workflow, examination…
BioTAK: A Biomarker Score for Earlier Takotsubo Syndrome Diagnosis
The five-item BioTAK score distinguished Takotsubo syndrome from acute coronary syndrome before invasive coronary angiography, achieving excellent discrimination in both development and external-validation cohorts.
Beyond Accuracy: A Five-Phase Path to Trustworthy Medical AI
Strong retrospective performance is only the beginning of medical AI evaluation. A new five-phase framework sets out the evidence needed to move diagnostic and predictive systems from technical…
Geriatrics & Aging: Biological Clocks Advance Earlier, Fairer Risk Detection
Recent Geriatrics & Aging coverage points toward multiscale aging clocks, low-burden risk markers, and fairness-aware dementia tools. The field is gaining earlier signals, but clinical utility still depends…
