Resource-Conscious Modeling for Next- Day Discharge Prediction Using Clinical Notes

Develop a template repository that benchmarks traditional ML (XGBoost) against small LLMs for specific clinical prediction tasks. Provide pre-built pipelines for data cleaning and tokenization of unstructured clinical notes.

Suggested repo: medpredict-light

"Benchmarking clinical prediction: BERT vs. XGBoost on real-world notes."

Estimated effort: 20h