112-LB: Proof-of-Concept Testing of an Artificial Intelligence–Based Fully Closed-Loop System in Hospitalized Patients with Diabetes

Objective: Achieving tight glucose control safely and effectively is challenging in the Intensive Care Unit (ICU). We conducted the first-in-human test of an artificial intelligence (AI)-based fully autonomous glucose control system in a simulated ICU setting. Methods: We admitted seven participants with type1 and type 2 (n=6) diabetes mellitus for 24-hour closed-loop glucose control sessions that included three unannounced meals. The AI-based system uses glucose control software that controls intravenous insulin and glucose infusions based on real-time continuous glucose monitoring derived data from two real-time CGM devices. The primary efficacy endpoint was the percentage of glucose values within the range of 70-180 mg/dL. The primary safety endpoint was the percentage of glucose values < 70 mg/dL. Results: Participants were on average 48±9.7 years-old, 71% female, and BMI 36±7.8 m/kg2. Mean time in the target range (70-180 mg/dl) was 87.8%. Mean time in the hypoglycemic range (< 70 mg/dL) was 0.2%, Figure. No adverse events occurred. Conclusion: This first in-human results show this novel fully autonomous closed-loop glucose control system designed for the ICU setting has the potential to achieve unparalleled glycemic control with minimal risk of hypoglycemia and paves the way for further research in the actual ICU setting.