Evaluating Machine Learning-Based Risk Labels in Electronic Health Records to Improve Prediabetes Understanding for Adults 65 and Older

What this Trial Is About

Researchers are evaluating whether adding machine learning-based risk information to electronic health record EHR lab result messages can help older adults aged 65 and older better understand their risk of developing prediabetes. The study focuses on adults with hemoglobin A1c levels in the range of 5.7% to less than 6.0%. This approach aims to improve patient understanding, emotional responses, quality of life, and healthcare use by providing personalized risk labels based on a predictive model developed from UCLA Health EHR data. Participants are automatically identified during lab result processing and randomly assigned to receive either standard lab result messages or modified messages that include a very low risk label generated by a machine learning model. The intervention uses a behavioral tool integrated into the EHR patient portal MyChart and provider interface, providing simplified messaging and context about current results. Both groups are followed in a real-world setting, with no updates made to the locked predictive model during the study. Participants complete two surveys one shortly after receiving their lab results and another about 30 days later, assessing understanding, emotional reactions, quality of life, and healthcare burden. The study also uses de-identified EHR data to track healthcare utilization, progression to diabetes, and provider responses over time. The total study period extends through 2029, with primary healthcare outcomes measured up to one year after lab results.

Study EHR Risk Stratification Tools