For people living with Type 1 Diabetes (T1D), keeping blood sugar levels in check is a constant challenge. A new clinical trial at UVA is aiming to simplify diabetes management by testing an innovative AI-powered device designed to improve automated insulin delivery.
The trial is co-led by several School of Data Science faculty, including Assistant Professor of Data Science Heman Shakeri ; Boris Kovatchev , founding director of the UVA Center for Diabetes Technology, a professor at the School of Medicine and professor of data science (by courtesy); and Anas El Fathi , research assistant professor at the Center for Diabetes Technology and assistant professor of data science (by courtesy).
Thanks to recent FDA approval, researchers are set to assess this breakthrough technology. The trial, kicking off in March, will evaluate a new reinforcement-learning feature called the "Bolus Priming System with Reinforcement Learning" (BPS_RL). Assisting with the fine-tuning of the BPS_RL was postdoctoral researcher Ali Tavasoli, who also generated the computer simulation that helped lead to the FDA approval.
The fully automated BPS_RL technology integrates with the existing Automated Insulin Delivery Adaptive NETwork (AIDANET) — a system comprising a phone app, Dexcom glucose monitor, and Tandem insulin pump — to enable insulin delivery without requiring user input.
The goal? To see if this new feature can help people maintain better blood sugar control, particularly during meals and overnight, while maintaining health safety and improving ease of use.
Addressing the Challenges of T1D
For many with T1D, keeping blood sugar levels stable is an ongoing struggle. Insulin needs can change due to meals, activity levels, stress, and other factors, making it difficult to dose accurately. AID systems require user input, and they can be costly and difficult to access. UVA's new study hopes to tackle these issues by developing a smarter, more adaptive system that is not only effective but also practical and affordable.
How It Works
Over the course of three weeks, 16 adult participants with experience using an AID system will test the enhanced technology.
Week 1: Participants use the standard AIDANET system at home to establish a baseline.
Week 2: Participants stay at a supervised testing location, using both the standard and updated systems for 18-hour sessions each.
Week 3: Participants return home and use the enhanced system under remote monitoring.
The study will compare how well blood sugar levels are maintained with and without the AI-powered upgrade. Half of the participants will start with the current system before switching to the new one, and the other half will do the reverse.
A New Era of Diabetes Care
"This trial isn't just about advancing technology — it's a bold step toward transforming diabetes care and uplifting lives," said Shakeri. "We are committed to creating a fully automated, intelligent insulin delivery system that redefines diabetes management, making treatment simpler, more reliable, and entirely effortless for patients."
Beyond improving blood sugar control, researchers hope that advancements like BPS_RL will help reduce the mental and financial burden of diabetes management. By making insulin delivery systems more adaptive, precise, and cost-effective, UVA is paving the way for a future where diabetes care is more efficient and equitable for all.
