1933-P: Digital Phenotyping of CGM Engagement Reveals Distinct Glycemic Outcomes
Diabetes · 05.06.2026 · 75 (Supplement_1)
DOI: 10.2337/db26-1933-p ↗Kurz gefasst
Fachartikel in der Zeitschrift Diabetes, erschienen am 05.06.2026. Verfasst von BOWEN ZHANG und 7 weiteren Autor:innen.
Abstract (Originalsprache)
Introduction and Objective: In routine practice, CGM adherence is often summarized as % days worn, ignoring whether wear is continuous or intermittent. We assessed whether longitudinal CGM wear patterns are associated with 12-month changes in CGM outcomes.
Methods
One-year CGM readings were linked to health records (EHR) for 2,351 U.S. Veterans. We quantified similarity between year-long CGM wear patterns with a novel optimal-transport-based metric that accounts for timing, gaps, and pattern complexity. We used spectral clustering to identify use phenotypes. Using double machine learning with EHR covariates, we estimated adjusted 12-month changes in time-in-range (TIR 70-180 mg/dL).
Results
Three phenotypes were identified: Consistent, Fluctuating, and Low-engagement. Compared with Consistent users, Fluctuating users had worse glycemic outcomes (change in TIR -3.6 percentage points), yet 73.2% met the conventional >=80% days-covered threshold. There was dose-response across use patterns, with Low-users deteriorating the most (change in TIR -7.0 points). Consistent wear related to larger favorable changes in TIR over 12 months among insulin pump users and those prescribed glucagon (Fig).
Conclusion
Glycemic benefit from real-world CGM is tied to CGM use patterns that are not captured by standard time of wear assessments. These results highlight the need for more sophisticated CGM use assessment and of tailoring patient education to improve individualized adherence goals. Disclosure B. Zhang: None. T. Okuno: None. E.M. Everett: None. D.R. Miller: None. G. Norman: Employee; Current; Dexcom, Inc. Stock/Shareholder; Current; Dexcom, Inc. P. Reaven: Research Support; Current; Dexcom, Inc., Lilly Global Health Partnership. H. Zhou: None. J. Zhou: None. Funding National Institutes of Health (GM141798), National Institutes of Health (DK142026), National Science Foundation (DMS-2054253), National Science Foundation (IIS-2205441), VA Merit
Zitieren
APA
ZHANG, B., OKUNO, T., EVERETT, E. M., MILLER, D. R., NORMAN, G., REAVEN, P., ZHOU, H., & ZHOU, J. (2026). 1933-P: Digital Phenotyping of CGM Engagement Reveals Distinct Glycemic Outcomes. Diabetes, 75(Supplement_1). https://doi.org/10.2337/db26-1933-p
Vancouver
ZHANG B, OKUNO T, EVERETT EM, MILLER DR, NORMAN G, REAVEN P, et al. 1933-P: Digital Phenotyping of CGM Engagement Reveals Distinct Glycemic Outcomes. Diabetes. 2026;75(Supplement_1). doi:10.2337/db26-1933-p
BibTeX
@article{ZHANG2026,
author = {ZHANG, BOWEN and OKUNO, TOMOKI and EVERETT, ESTELLE M. and MILLER, DONALD R. and NORMAN, GREG and REAVEN, PETER and ZHOU, HUA and ZHOU, JIN},
title = {1933-P: Digital Phenotyping of CGM Engagement Reveals Distinct Glycemic Outcomes},
journal = {Diabetes},
year = {2026},
volume = {75},
number = {Supplement_1},
doi = {10.2337/db26-1933-p}
}
EUVIMED