A machine learning pipeline that predicts the probability of a patient not attending ("DNA") an outpatient appointment at Barts Health NHS Trust. Around 12% of outpatient appointments are missed each month, costing thousands of hours of clinical time; identifying high-risk appointments in advance lets the Trust target support before the appointment date.
The pipeline trains a per-site XGBoost classifier on historic appointment, demographic, and clinical data, scores future appointments, and surfaces results through a Streamlit dashboard.
- Architecture — pipeline stages and data flow
- Setup — installation, configuration, and running locally
See docs/setup.md.
Data/, Results/, Research/, Old_Scripts/, and Documentation/ are
gitignored and hold private Trust data and internal project material — they
are not published.