The Clinical Excellence Research Center is exploring applications of machine learning to electronic health record data and to administrative claims data. These efforts use machine learning to provide powerful insights like the identification of patients likely to incur high medical costs in future time periods.
Projects and Outcomes
CERC conducted a population-based study of newly high-cost patients using the Danish national database. We used machine learning to predict future high-cost patients by identifying those whose costs will "bloom" within 12 months. Our analysis found that we improved positive predictive power by more than 30%, compared with standard risk-stratification tools.