Glycemic Health
DiabetesWatch
Our lab is developing digital biomarkers of glycemic health from wearables data physiology, with a focus on robustness across diverse glycemic variability, interstitial glucose ranges, and hemoglobin A1c values. Building on preliminary algorithms, we are validating performance and extending methods so they translate from research-grade wearables to consumer smartwatches. This work blends digital health analytics with physiology, time-series modeling, and practical data engineering. Students and collaborators can contribute to study design, wearable data preprocessing, feature development, model evaluation, and deployment-oriented validation that aligns with our broader wearable biomarker research.
Funded by: NIH R01 DK133531
Related Publications: Bent, Brinnae, Peter J. Cho, Maria Henriquez, April Wittmann, Connie Thacker, Mark Feinglos, Matthew J. Crowley, and Jessilyn P. Dunn. “Engineering digital biomarkers of interstitial glucose from noninvasive smartwatches.” NPJ Digit Med 4, no. 1 (June 2, 2021): 89. https://doi.org/10.1038/s41746-021-00465-w
Lab members: Karnika Singh, Leeor Hershkovich, Shekh md mahmudul Islam, Bill Chen, Hayoung Jeong, Keying Jia
Collaborators: Dr. Anastasia-Stefania Alexopoulos, Dr. Matthew Janik Crowley
Glycemic Profiling and Phenome-wide Associations in the AI-READI Dataset
Our lab is leveraging the AI-READI dataset to study glucose dynamics using continuous glucose monitoring (CGM) alongside multimodal phenotyping. The cohort includes 1,067 participants spanning healthy, prediabetic, and type 2 diabetic statuses, with clinical biomarkers and wearable-derived lifestyle measures available for integrative analysis. We are investigating associations between CGM-derived glycemic patterns and clinical, behavioral, and physiological variables both across and within glycemic strata to identify new risk factors and actionable markers. Students and collaborators can contribute to CGM preprocessing, feature engineering, multimodal data fusion, statistical modeling, and reproducible analytics workflows.
Lab members: Bill Chen, Annika Kumar
Collaborators: Dr. Anastasia-Stefania Alexopoulos
LeMonAids: Leveraging multimodal Monitoring to Ascertain Intra-individual Diabetes Subphenotypes
Our lab is developing LeMonAids, a multimodal foundation model to better characterize glycemic variability in type 2 diabetes and uncover intra-individual subphenotypes that can guide personalized management. We are extending the GluFormer glucose prediction embedding space by integrating smartwatch physical activity signals with CGM and diet logs, training with next-token prediction to capture lifestyle-driven dynamics in glucose responses. We will validate postprandial glycemic response prediction on external cohorts (AI-READI, Diabetes Watch) and build an interactive, open-use case for patients and clinicians via community platforms. Students and collaborators can contribute to data fusion, model training, evaluation, and tool-building.
Lab members: Leeor Hershkovich, Bill Chen
Collaborators: Dr. Anastasia-Stefania Alexopoulos
