Biobehavioral Human-Machine Co-adaptation of the Artificial Pancreas

点击翻译按钮获取中文摘要

PROJECT SUMMARY Biobehavioral Human-Machine Co-adaptation of the Artificial Pancreas Closed-loop control (CLC) is now transitioning to the clinical practice and one of the most advanced systems to date–Control-IQ–uses an algorithm designed and tested by the previous research cycle of this project. With the first generation of our CLC system now translated to the clinic, our objective is to design and test next-generation CLC solutions, learning from the experience and utilizing the large database accumulated to date. Thus, we focus this project on the new concept of Adaptive Biobehavioral Control (ABC) – a first-in-class system that will use human-machine co-adaptation of CLC, recognizing both the necessity for the control algorithm to adapt to changes in human physiology, and the necessity for the person to adapt to CLC action. To achieve its objectives, the ABC system will have two new components added to the current state-of-the art Control-IQ: a Behavioral Adaptation Module (BAM) – a behavioral intervention deployed in a mobile app to assist a person's adaptation to CLC by information and risk assessment primarily regarding meals and physical activity, and a Physiologic Adaptation Module (PAM) – an automated procedure tracking risk status and changes in the user's metabolic profile and acting in real time to adapt the CLC algorithm's insulin control parameters. Using these technologies, we now propose to compare, in a randomized cross-over trial enrolling 90 participants with type 1 diabetes, the current CLC (Control-IQ) to three new treatment modalities: ABC and its components BAM and PAM. To do so, study participants will be randomized to two groups following two different sequences of treatment modalities: CLCCLC+BAMCLC+PAMABC and ABCCLC+PAMCLC+BAMCLC. Each treatment modality will continue for 2 months and the treatments will be separated by 2-week washout periods. This design was used successfully in our previous study and enables four crossover comparisons: CLC vs. ABC (primary) and CLC+BAM vs. CLC; CLC+PAM vs. ABC; CLC+BAM vs. CLC+PAM (secondary). We expect that: (1) ABC will be superior to the current CLC in terms of: improved time in the target range 70-180mg/dl measured by continuous glucose monitoring (CGM); reduced risk for hypoglycemia, and better technology acceptance; (2) Behavioral adaptation (CLC+BAM) will be superior to CLC in terms of improved CGM-measured time in the target range during the day and reduced CGM-measured incidence of hypoglycemia during/after exercise; (3) Physiologic adaptation (CLC+PAM) will account for most of the glycemic benefits of ABC overnight, will be inferior to BAM in terms of postprandial glucose variability and hypoglycemia during/after exercise, and will be superior to BAM in terms of technology acceptance for those who prefer fully-automated control. Overall, we affirm that reliable technology has been developed and sufficient data accumulated to warrant the development of next-generation biobehavioral control, aiming adaptation of user behavior to the specifics of CLC treatment and adaptation of CLC technology to user physiology, separately and in combination. The proposed project will design and test a system for monitoring and control of these adaptation processes.