EHR Risk Alternatives to opaque clinical AutoML
If you are searching for alternatives to black-box clinical AutoML for research and education, prefer tools that expose splits, leakage checks, calibration, and explanations—not only a leaderboard AUROC.
What to look for
- Explicit index time / prediction horizon
- Temporal or group CV—not random row CV on longitudinal EHR
- Calibration metrics and plots
- Inspectable models + SHAP (or similar)
- Clear “not a medical device” framing
EHR Risk Framework as an open alternative
An open-source framework for leakage-safe, calibrated, and explainable EHR risk prediction— Docker workbench, FastAPI jobs, light research HPO (not clinical AutoML), and downloadable reports.