The Study

The GLOW UP study aims to evaluate the predictive potential of wearable-based lifestyle factors—such as physical activity, sleep, and diet—in the prevention of type 2 diabetes (T2D) [1-6]. Specifically, it focuses on identifying prediabetes using fasting plasma glucose (FPG) and HbA1c measurements [7]. By leveraging real-world data from commercially available wearables, this study investigates whether lifestyle factors, continuously monitored in free-living conditions, can effectively predict early abnormalities in glucose metabolism. Moreover, it seeks to determine the relative importance of each lifestyle factor, offering insights into personalized glucose dynamics and lifestyle patterns, which could further advance the field of personalized T2D prevention.

This prospective case-control study involves a four-week period of continuous lifestyle monitoring using wearable devices. Participants at risk are adults aged 45 years and older, with a body mass index (BMI) exceeding 25 kg/m² [8], all residing in Switzerland and selected based on predefined criteria. A total of 200 participants will be enrolled, comprising 100 healthy individuals and 100 individuals diagnosed with prediabetes. Lifestyle behaviors—including physical activity, sleep, diet, stress, and demographic factors—will be tracked using fitness trackers and smartphone applications. To monitor glucose levels, blinded continuous glucose monitoring (CGM) will be employed. Both baseline and follow-up measurements will include HbA1c, FPG, and anthropometric parameters. Predictive modeling techniques, including classification and regression-based models, will be used to evaluate the relationship between lifestyle factors and glucose metrics. Additionally, clustering and regression analyses will explore the association between glucose profiles and various metabolic characteristics.

The primary outcome of the study is to assess the predictive performance of wearable-derived lifestyle data in categorizing HbA1c and FPG values. Secondary outcomes focus on the ability of wearable data to predict time-varying CGM-based metrics, such as time-in-range, glucose variability, glucodensity profiles, and postprandial glucose excursions. Further analyses will explore the interaction between lifestyle patterns (e.g., heart rate variability, stress, nutrition, physical activity, and sleep) and CGM-based metrics across diverse metabolic phenotypes, defined based on FPG, HbA1c, BMI, and visceral fat measurements. This approach aims to reveal how lifestyle patterns and glucose dynamics vary across different metabolic profiles.

The GLOW UP study has the potential to significantly contribute to the fields of digital health and precision medicine. By introducing a data-driven approach to T2D prevention using digital biomarkers, the study’s findings are expected to support the development of personalized, data-driven strategies for early T2D risk assessment and preventive healthcare, specifically within the Swiss population.

Objectives

The overall objective of the GLOW UP study is to investigate whether, and to what degree, smartphone and wearable-based lifestyle measures can (a) accurately detect prediabetes in free-living conditions, and (b) explain and predict heterogeneity in metabolic profiles among adults at-risk of T2D.

1. The primary objective of this study is to assess the accuracy and performance of smartphone and wearable-based lifestyle metrics, along with baseline demographics, in predicting HbA1c and fasting plasma glucose (FPG) classifications. These classifications distinguish between healthy individuals and those with prediabetes, as defined by gold-standard diagnostic thresholds. Specifically, HbA1c levels below 5.7% and FPG levels below 5.6 mmol/L are classified as healthy, while HbA1c levels between 5.7% and 6.4% and FPG levels between 5.6 and 6.9 mmol/L indicate prediabetes.
2. At the within-person level, we aim to evaluate the relative contribution of different lifestyle measures (e.g., nutrition, physical activity, and stress) in predicting fluctuations in time-varying CGM-based metrics such as postprandial glucose excursions.  
3. At the between-person level, we aim to evaluate individual differences in lifestyle patterns (e.g., HRV, stress, nutrition, physical activity, and sleep) and CGM-based metrics (time-in range, glucose variability, and glucodensity profiles vary across metabolic phenotypes (derived based on FPG, HbA1c, BMI, and visceral fat).
4. We aim to assess the reliability and accuracy of image-based dietary recognition algorithms in predicting nutrition compositions compared to traditional methods, such as trained annotator evaluations.
5. We aim to investigate how dynamic notification reminders (e.g., personalized or incentive-driven prompts) and lifestyle factors (e.g., momentary stress) influence adherence to image-based meal tracking.

[1]   A. G. Tabák, C. Herder, W. Rathmann, E. J. Brunner, and M. Kivimäki, “Prediabetes: a high-risk state for diabetes development,” Lancet, vol. 379, no. 9833, pp. 2279–2290, Jun. 2012.

[2]   Diabetes Prevention Program Research Group, “10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study,” The Lancet, vol. 374, no. 9702, pp. 1677–1686, Nov. 2009.

[3]   “Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial,” The Lancet, vol. 368, no. 9541, pp. 1096–1105, Sep. 2006.

[4]   W. C. Knowler et al., “Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin,” N Engl J Med, vol. 346, no. 6, pp. 393–403, Feb. 2002.

[5]   A. Ramachandran, C. Snehalatha, S. Mary, B. Mukesh, A. D. Bhaskar, and V. Vijay, “The Indian Diabetes Prevention Programme shows that lifestyle modification and metformin prevent type 2 diabetes in Asian Indian subjects with impaired glucose tolerance (IDPP-1),” Diabetologia, vol. 49, no. 2, pp. 289–297, Jan. 2006.

[6]   J. S. Torgerson, J. Hauptman, M. N. Boldrin, and L. Sjöström, “XENical in the prevention of diabetes in obese subjects (XENDOS) study. A randomized study of orlistat as an adjunct to lifestyle changes for the prevention of type 2 diabetes in obese patients,” Clinical Diabetology, vol. 5, no. 2, pp. 95–104, 2004.

[7]   “Diabetes Diagnosis & Tests.” Accessed: Feb. 19, 2025. [Online]. Available: https://diabetes.org/about-diabetes/diagnosis

[8]   “Risk Factors for Diabetes,” National Institute of Diabetes and Digestive and Kidney Diseases. Accessed: Apr. 27, 2025. [Online]. Available: https://www.niddk.nih.gov/health-information/professionals/clinical-tools-patient-management/diabetes/game-plan-preventing-type-2-diabetes/prediabetes-screening-how-why/risk-factors-diabetes

The GLOW UP study assesses the predictive value of wearable-based lifestyle factors (e.g., physical activity, sleep, diet) for preventing type 2 diabetes. By leveraging real-world data from commercially available wearables, the study explores whether lifestyle factors monitored in free-living conditions can predict early glucose metabolism abnormalities.

Dr. Mia Jovanova, Victoria Brügger, Magdalena Fuchs, Qiuhan Jin, Prof. Dr. Tobias Kowatsch

Prof. Dr. med. Michael Brändle

Dr. med. Benjamin Wirth

September 2025 – TBD

This project is funded by the CSS and is part of the CSS Health Lab.