-
1
The collection’s 31 contributions span video, physiological signals, digital health, and population health, emphasizing interpretable systems that integrate domain knowledge.
-
2
Portable video models support assessment of gait, autism, ophthalmologic conditions, ultrasound lesions, blood counts, and cardiovascular risk beyond laboratory settings.
-
3
Signal-analysis applications use lightweight and multimodal models to predict outcomes in neurology, psychiatry, and cardiology, including from ECGs and routine records.
-
4
Combining wearables, biomarkers, multi-omics, and electronic health records captures health complexity across applications such as Alzheimer disease, cancer, and frailty.
-
5
The authors identify mechanistic interpretability, human-centered explanations, prospective validation, adaptive regulation, and equitable deployment as priorities for clinical translation.
