JMIR Publications

ABSTRACT

Background:

The growth in capabilities of telehealth have made it possible to identify individuals with a higher risk of uncontrolled diabetes and provide them with targeted support and resources to help them manage their condition. Thus, predictive modeling has emerged as a valuable tool for the advancement of diabetes management.

Objective:

The objective of this study was to conceptualize and develop a novel machine learning approach to proactively identify members enrolled in a remote diabetes monitoring program (RDMP) who were at-risk of uncontrolled diabetes at 12 months in program. Therefore, a set of dynamic predictive machine learning models were designed and trained at specific check points during the member’s time in program to proactively identify those at-risk and capture member attributes that could impact the member’s at-risk status.

Methods:

Registry data from the RDMP was used to design separate machine learning models to predict member outcomes at each monthly checkpoint of their program journey (month-n models) from the first day of onboarding (month-0 model) up to the 11th month (month-11 model). A member’s program journey began upon onboarding into the RDMP and self-monitoring blood glucose (SMBG) through the program blood glucose (BG) meter. Each member passed through 12 predicative models through their first year enrolled in the RDMP. Four categories of member attributes, including survey data, BG data, medication fills, and health signals, were used for feature construction. Models were trained using LightGBM and underwent hyperparameter tuning. The performance of the models was evaluated using standard metrics, including precision, recall, specificity, AUC, F1 score, and accuracy.

Results:

The models exhibited strong performance, accurately identifying observable at-risk members with a recall ranging from 70% to 94% and precision from 40% to 88% across the 12-month program journey. Unobservable at-risk members also showed promising performance with recall ranging from 61% to 82% and precision from 42% to 61%. Overall, model performance improved as members progressed through their program journey, demonstrating the importance of engagement data in predicting long-term clinical outcomes.

Conclusions:

Members’ temporal and static attributes, identification of diabetes management patterns and characteristics, and their relationship to predict the diabetes management outcome was explored. Proactive targeting models accurately identified members at-risk for uncontrolled diabetes with a high level of precision that was generalizable through future years within the RDMP. The ability to identify members who are at risk at various time points throughout the program journey allows for personalized interventions to improve outcomes. This approach offers significant advancements in the feasibility of large-scale implementation in remote monitoring programs and can help prevent uncontrolled glycemic levels and diabetes-related complications. Future research should include the impact of significant changes that can affect a member’s diabetes management.