Using Exercise to Inform Insulin Administration in Patients With Type 1 Diabetes

THANKS. Listen to this article using the player above. ✖

Around half a billion people worldwide are diagnosed with diabetes. A studyled by Dr. Peter G. Jacobs at Oregon Health and Science University and published in Lancet digital healthdemonstrates the potential of integrating data from wearable fitness sensors and artificial intelligence (AI) to inform insulin delivery systems on when to administer insulin after physical activity.

The current lack of blood sugar regulation during exercise

Type 1 diabetes is an autoimmune disease that causes the loss of insulin-producing beta cells in the pancreas, leading to dysregulation of blood sugar levels. Historically, diabetes has been treated with insulin injections or the use of an insulin pump. However, automated insulin delivery (AID) systems have become more popular recently as a method of insulin delivery because they contain a glucometer that can automate insulin delivery based on monitor readings and needs. in glucose from the body. Although AIDs have been shown to be helpful, they still need improvement because patients can often experience low blood sugar levels even though they are under the control of an AID system. One of the main reasons for this is that there are more physiological factors to consider when a person exercises, making the required response to AIDS less predictable during these times. To address this issue, Jacobs et al. wanted to see if AI could be used to analyze data from wrist-worn fitness monitors, which could then be used to inform AIDS. In this study, they tested two AI algorithms: exercise-sensitive model predictive control (exMPC) and exercise-sensitive adaptive proportional derivative (exAPD). The exMPC has the ability to detect major changes in physical activity, such as during a run, and the exAPD has the ability to detect similar activities as well as low-intensity activities, such as household chores.

AI is effective at integrating physical activity and blood sugar information

This study was conducted at the Harold Schnitzer Diabetes Health Center at Oregon Health and Science University and included 25 participants. All participants had been diagnosed with type 1 diabetes for at least 1 year and were aged 21 to 50 years. First, patients underwent a one-week “run-in” period, during which their glucose levels were monitored but without an algorithm. was used to modulate the activity of the AID system. After this period, patients began a treatment regimen with implementation of the algorithms, during which they participated in a 76-hour session under the control of an algorithm, and then in another 76-hour session under the control of the other algorithm. During each session, they were first followed in the research center and then at home, during which they performed activities of daily living, such as doing laundry or washing dishes, as well as an activity more intense physical, like timed workouts, the whole thing. which was captured via the wrist-worn fitness watch. Throughout the study, researchers recorded participants’ blood sugar levels via the Dexcom G6 CGM.

The main conclusions of the paper were:

• There was no significant difference between participants under the exMPC algorithm and participants under the exAPD algorithm with respect to the percentage of time spent below the standard blood glucose range or the time spent in the range when of the first primary session in the clinic.
• During the 76-hour sessions in which exercise was introduced, the two algorithms, exMPC and exAPD, performed similarly in terms of time in range (71.2% and 75.5% respectively) and time below the range (0.96% and 1.30% respectively).
• Both algorithms were more successful in minimizing time below range for the duration of the test, compared to when no algorithm was used, which had a time below range of 2.4 %.

Using AI and Wearable Technology in Patients With Type 1 Diabetes

Current blood glucose monitoring and insulin delivery systems for patients with type 1 diabetes can be unpredictable in their response after exercise. The findings of Jacobs et al. suggest that AI has the potential to improve the care of patients with type 1 diabetes, which includes real-time monitoring of patients’ glucose levels, particularly after physical activity. The two algorithms used in the study, exMPC and exAPD, show potential for integration into existing AID systems, as they both provide improved ability to maintain glucose levels within range and minimize time spent in the beach. Participants spent less time under the scope when monitored by one of the algorithms, compared to no algorithm at all. Study results indicate that the use of exercise measurements, collected via wearable technology, can be used by both exMPC and exAPD to inform insulin delivery in AIDS.

Although this study presents promising results, some limitations should be addressed in future research. First, the study size was small and participants were followed over a relatively short period of time. Additionally, exercise sessions within the clinic were limited to aerobic exercise, rather than other forms of exercise. Finally, the study population consisted of people whose blood sugar levels were normally well monitored, which is not representative of all people with type 1 diabetes.

Next steps to continue testing their model

Future research on this topic should include longitudinal studies that monitor participants over a longer period of time to understand the impact of algorithm implementation on their long-term health. For statistical power, a larger cohort of patients is also needed. Additionally, future studies should use other types of exercise, in addition to only aerobic exercise, to account for the diversity of activities participants may undertake in normal life. This study demonstrated the potential of AI to improve current AID systems to optimize insulin delivery to patients, particularly when exercising, and future research should continue to build on on the basis established by this study.

Reference: Jacobs PG, Resalat N, Hilts W, et al. Integrating metabolic expenditure information from wearable fitness sensors into an AI-augmented automated insulin delivery system: a randomized clinical trial. Health of the Lancet figures. 2023;5(9):e607-e617. do I:10.1016/S2589-7500(23)00112-7