(P15-017-23) Deep Profiling of Diet, Sleep, and Physical Activity Using Wearable Biosensors Reveals Associations Between Lifestyle Timing and Glucose Levels

Objectives: Understanding the relationship between glucose levels and lifestyle at a personal level is crucial for the management of Type 2 diabetes (T2D). Modern technologies such as wearable sensors and smartphone applications have enabled 24-hour monitoring of lifestyle behaviors in real time. Using continuous glucose monitoring (CGM) and wearables, this study searched for associations between lifestyle features and glucose levels at high resolution.

Methods: 36 study participants (56.1y; prediabetes/T2D n=20 and normoglycemia n=16) tracked their food intake and timing by a food-logging mobile app, and sleep and activity by a Fitbit band for at least two weeks. Gold-standard glucose metabolic tests were conducted including OGTT, insulin suppression test, and isoglycemic intravenous glucose infusion to quantify muscle insulin resistance, beta cell function, and incretin function. CGM was used to monitor glucose fluctuation throughout the study.

Results: Through the collection of lifestyle data (2,307 meals, 1,809 days for sleep, and 2,447 days for activity) and 231,206 CGM readings from a well-phenotyped set of individuals (n=36), we found using machine learning that: 1) Meal timing (the caloric contribution from different time windows) was associated with HbA1c, fasting plasma glucose, beta cell function, incretin function, and CGM (P<0.036); 2) Sleep timing was associated with CGM and incretin function (P<0.029); 3) Timing of increased activity showed differential effects: CGM-lowering benefits from increasing afternoon activity in IS group whereas morning activity benefited the IR (P < 0.001); and 4) Network analysis revealed that diet, sleep, and activity are tightly interrelated (P < 0.05). Notably, dietary features including eating timing were central in the network.

Conclusions: Our data demonstrate that diet, sleep, and physical activity are closely interrelated, and they are strongly associated with divergent glucose outcomes measured objectively by CGM and standardized metabolic tests. We also found obvious diurnal patterns in lifestyle behaviors that influence glucose physiology and implicate the timing of food intake, sleep, and exercise could be a powerful behavioral regimen to ensure appropriate glycemic control.

Funding Sources: NIH-2T32HL09804911, NIH-5R01DK110186-02, Stanford PHIND Award, Stanford Lifestyle Medicine Award