Hospital readmission predictor

The brief

Patients readmitted soon after discharge cost hospitals money and often signal gaps in care. Predict which diabetic patients are likely to be readmitted within 30 days, handle the imbalance properly, and explain each patient's risk so nurses can plan follow-up calls. Good looks like: careful feature work on medical codes, metrics for an imbalanced outcome, and explanations a care team accepts.

What you will build

1. Clean the encounters and group diagnosis codes
2. Define 30-day readmission and avoid leakage
3. Train gradient boosting with imbalance handling
4. Evaluate with recall, precision and PR-AUC
5. Explain risk for individual patients with SHAP

Free, open data to build it with

• Diabetes 130-US hospitals for years 1999–2008 ↗ UCI Machine Learning Repository

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.