Managing blood glucose levels during and after exercise is a critical challenge for individuals with type 1 diabetes. Thanks to the integration of wearable fitness sensors, a recent clinical study demonstrated the potential of automated insulin delivery (AID) systems to adjust insulin dosing based on real-time fitness data. This advancement holds the promise of reducing hypoglycemia and enhancing glucose outcomes during physical activity and daily life.


Type 1 diabetes, an autoimmune disorder, impairs the pancreas’ ability to produce insulin, which is essential for regulating blood glucose concentrations. People with this condition often require exogenous insulin to maintain normal glucose levels. Automated insulin delivery (AID) systems have gained traction in recent years, comprising a combination of continuous glucose monitoring (CGM), insulin pumps, and control algorithms. These systems automate insulin delivery based on sensed glucose levels, but there’s a persistent need for improvement, as achieving target glucose levels and avoiding hypoglycemia remains a challenge for many.


The study, published on The Lancet Digital Health and conducted at the Harold Schnitzer Diabetes Health Center clinic at Oregon Health and Science University, focused on harnessing wearable fitness sensors to enhance AID systems’ performance. Two distinct AID algorithms were evaluated: the exercise-aware adaptive proportional derivative (exAPD) algorithm and the exercise-aware model predictive control (exMPC) algorithm. Both algorithms integrated real-time fitness data from smartwatches to inform insulin dosing and counter hypoglycemia during exercise and daily activities.

27 participants with type 1 diabetes were enrolled in the study. These participants underwent a crossover design, where they experienced both exAPD and exMPC algorithms in different sessions. The primary outcome measured was time spent below a certain glucose range (<3.9 mmol/L) during the primary in-clinic session. The secondary outcomes included mean glucose levels and time spent in the target glucose range (3.9–10 mmol/L).


The results indicated that there was no significant difference between the exMPC and exAPD algorithms in terms of time spent below the target glucose range or time spent in range during the primary in-clinic session. However, in the two-hour period following the start of in-clinic exercise, the exMPC algorithm demonstrated lower mean glucose levels compared to the exAPD algorithm. Across the entire 76-hour study duration, both algorithms successfully achieved clinical time in range targets, with significantly lower time spent below range compared to the run-in period.

This study’s outcomes indicate a step forward in optimizing automated insulin delivery for individuals with type 1 diabetes. By incorporating wearable fitness sensor data, AID systems show potential for enhancing glucose outcomes and reducing hypoglycemia risks during exercise and daily activities. The integration of exercise metrics from wearable fitness sensors into future AID systems could open doors to safer and more effective exercise management for those living with type 1 diabetes.

Funding for this study was provided by the JDRF Foundation, the Leona M and Harry B Helmsley Charitable Trust, and the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. The findings highlight ongoing advancements in health technology that aim to improve the quality of life for individuals with chronic conditions.