Researchers from the DQBM Krauthammer Lab developed machine learning models to identify patients at high risk of missing MRI appointments at University Hospital Zurich. Using more than 38,000 appointments, XGBoost performed best and was subsequently evaluated prospectively in routine practice.
A nine-month trial then tested reminder calls for high-risk patients. The calls did not significantly reduce the overall no-show rate, although patients who were reached attended more often than those who were not. The study also showed how changes in clinical data over time can reduce model performance, emphasizing the importance of continuous monitoring and adaptive interventions.
Publication: DOI 10.1093/jamiaopen/ooag151.