The University of Virginia (UVA) is embarking on a groundbreaking clinical trial to assess an innovative AI-powered device aimed at improving diabetes management for people with Type 1 Diabetes (T1D). The new technology, designed to enhance automated insulin delivery, is being tested through a partnership between several experts at UVA’s School of Data Science, including Heman Shakeri, assistant professor; Boris Kovatchev, founding director of the UVA Center for Diabetes Technology; and Anas El Fathi, research assistant professor at the same center.
Following FDA approval, the trial will evaluate the device’s performance, specifically the newly developed “Bolus Priming System with Reinforcement Learning” (BPS_RL). This feature, which leverages reinforcement learning, was fine-tuned with the help of postdoctoral researcher Ali Tavasoli. The BPS_RL is integrated with the existing Automated Insulin Delivery Adaptive NETwork (AIDANET), a system that combines a smartphone app, a Dexcom glucose monitor, and a Tandem insulin pump to automate insulin delivery without requiring user input.
The trial, set to begin in March 2025, aims to test whether this advanced AI technology can improve blood sugar control, particularly during meals and overnight, while simplifying usage and enhancing safety for patients. The researchers hope to create a system that can adjust more seamlessly to the daily fluctuations in insulin needs, which are influenced by factors such as meal timing, physical activity, and stress levels.
For many individuals with Type 1 Diabetes, achieving stable blood sugar levels is a persistent challenge due to the complex and varying factors that impact insulin requirements. Traditional AID systems still rely on user input and can be costly and difficult to access. UVA’s clinical trial is seeking to address these concerns by developing a more adaptive, reliable, and cost-effective solution for diabetes care.
During the trial, 16 adult participants who are familiar with using AID systems will evaluate both the current system and the updated, AI-enhanced version over a three-week period. In the first week, participants will use the standard AIDANET system at home. In the second week, they will stay at a supervised facility, alternating between the standard and upgraded systems in 18-hour testing sessions. Finally, during week three, participants will return home and continue using the enhanced system under remote monitoring.
Half of the participants will begin with the standard system before switching to the AI-powered version, while the other half will follow the reverse sequence. By comparing how each version impacts blood sugar control, researchers aim to assess the effectiveness of the AI-enhanced system.
“This trial represents more than just a technological advancement—it’s a pivotal step in transforming the landscape of diabetes care,” said Shakeri. “Our goal is to create an intelligent, fully automated insulin delivery system that simplifies treatment and improves the overall quality of life for people living with Type 1 Diabetes.”
In addition to improving blood sugar management, researchers hope that advancements like BPS_RL will help reduce the mental and financial strain often associated with diabetes care. By making insulin delivery systems more precise, adaptive, and accessible, UVA is working to ensure that diabetes care becomes more efficient and equitable for all.
Related topics:
SGLT2 and GLP-1 Diabetes Medications: Tailoring Treatment for Optimal Cardiovascular Protection
Blood Vessels Linked to Obesity-Driven Diabetes: Key Molecule Identified in New Research
Gut Microbiome Changes Linked to Higher Diabetes Risk from Sugary Drinks
