Analyzed 100K+ hospital encounter records to identify trends influencing 30-day readmissions across diagnoses, length of stay, discharge type, and patient history.

Performed data cleaning, cohort analysis, and exploratory analysis to surface drivers of readmission risk and follow-up demand.

Developed a risk scoring and forecasting workflow to help prioritize high-risk patients instead of treating predictions as black-box outputs.

Designed a RAG-based clinical reasoning layer to generate human-readable explanations for model outputs and built a weekly staffing and follow-up planning module.