“Can we identify patients at high risk for hospital readmission or adverse outcomes, so that care teams can intervene earlier?”

Tools Used: Logistic Regression, XGBoost, DataRobot (for classification & risk scoring)

Functionality:
Automated risk flagging within EHR workflow

Personalized risk stratification dashboards

Status of Data:

Working with synthetic Electronic Health Rrecords (EHR) data (Synthea)

Data prep includes demographics, comorbidities, prior admissions, lab values, and utilization history

Remaining Relevant Information:

Evaluation metrics: AUROC, AUPRC, calibration plots

Model interpretability: SHAP values for feature contributions

Impact: Reduce costly readmissions, improve patient safety and care quality