Healthcare analytics
Earlier insight. Better healthcare decisions.
My healthcare work spans disease risk, preventive intervention, financial sustainability, patient prioritization, hospital forecasting, and operational optimization.
Case Study 02
Early Diabetes Risk Prediction
Using public health indicators to identify elevated risk and support earlier intervention.
Business problem
Delayed identification
Healthcare organizations need practical methods to identify high-risk individuals before avoidable complications occur.
Approach
Classification modeling
Logistic Regression, Random Forest, and XGBoost evaluated with recall, precision, F1, and ROC-AUC.
Business value
Preventive outreach
Risk scores support targeted education, screening, and allocation of preventive resources.
Case Study 03
Predicting Preventable High-Cost Patients
Using prior-year data to identify future high-cost users before crisis care.
0.817Sealed-validation ROC-AUC
4.31×Top-decile lift
42.7%High-cost cases captured in top decile
Case Study 04
Hospital Resource Optimization
Forecasting next-week occupancy and translating predictions into proactive operational decisions.
| Model | Purpose | Outcome |
|---|---|---|
| Persistence baseline | Operational benchmark | Current occupancy carried forward |
| Linear Regression | Interpretable benchmark | Linear trend comparison |
| Random Forest | Nonlinear relationships | Strong out-of-sample performance |
| XGBoost | Selected model | Average R² 0.907; RMSE 2.888 |