Objective
We aim to develop and test Automated Bolus eXtension (ABX), a software upgrade for current insulin pumps that automatically delivers mealtime insulin—without the user needing to count carbohydrates, push buttons, or announce meals. ABX works by learning from each person’s real-world data: how their glucose responds to food, insulin, and daily routines. It continuously adapts to individual patterns and makes safe insulin dosing decisions throughout the day. We will test ABX in a clinical study with teens and young adults using commercially available insulin pumps, evaluating its ability to improve blood sugar control and ease the daily demands of living with diabetes.
Background Rationale
Even with today’s advanced “smart” insulin pumps, people with type 1 diabetes—especially teens and young adults—must still manually enter every meal and guess how many carbohydrates they are eating. This process is stressful, error-prone, and often skipped, leading to sharp spikes in blood sugar after meals and a higher risk of long-term complications. Despite recent advances, there is still no system that can fully automate meal-time insulin. Thanks to breakthroughs in artificial intelligence (AI), we can now build software that recognizes personal glucose patterns, learns daily habits, and safely predicts the right insulin dose—even without a meal announcement. By embedding this technology into existing insulin pumps, we aim to finally remove the most burdensome part of diabetes care: mealtime insulin dosing.
Description of Project
People living with type 1 diabetes (T1D) must think about insulin dozens of times every day, especially at meals. Over the past two decades, Automated Insulin Delivery (AID) using insulin pumps and glucose sensors have revolutionized care by automatically adjusting insulin and protecting users from low-glucose events. Yet today’s systems still depend on the person to estimate the carbohydrates in every snack or meal and press buttons to deliver an insulin “bolus” dose. That extra step is stressful, easy to forget, and a major reason why teenagers and young adults—who already face hormonal changes and busy schedules—have the most challenging glycemic control of any age group.
Our team at the UVA Center for Diabetes Technology, which helped develop the first commercial AID systems, now aims to remove this manual task. We are building an Automated Bolus eXtension (ABX)—a smart software that plugs into existing hybrid AID, learns each user’s unique glucose and lifestyle patterns, and delivers the right meal-time insulin automatically, without any carb counting or button-presses.
To show ABX works in real life, we will run a 20-week study with 48 adolescents and young adults. Participants will first compare their usual hybrid pump with and without ABX, then try an enhanced version (ABX+) that fine-tunes itself every month using advanced artificial-intelligence techniques. We will track the time their glucose stays in the healthy range, the number of low-blood-sugar episodes, and how much easier diabetes feels day-to-day.
By taking the math out of meal-time, ABX could give young people safer glucose levels, fewer scares, and the freedom to focus on life—not diabetes—while paving the way for fully hands-free diabetes technology for everyone.
Anticipated Outcome
We expect the ABX system to:
• Increase the number of hours each day that blood glucose levels stay in the healthy range (Time-in-Range);
• Reduce large swings in glucose after meals, including both high spikes and dangerous lows;
• Lighten the mental burden and anxiety that often come with diabetes self-management.
In our study, we will evaluate ABX in a real-world setting with young people who use insulin pumps. If successful, ABX could be made available as a software update—no new hardware required—and could transform current insulin pumps into fully automated systems, making diabetes care more effortless and consistent.
Relevance to T1D
Mealtime insulin dosing is one of the most difficult and stressful parts of managing type 1 diabetes. It’s also one of the remaining manual step in otherwise automated insulin delivery systems. For many young people, this step is skipped or delayed, increasing the risk of complications and making life with diabetes harder than it needs to be. ABX is designed to close that gap—making it possible to safely and automatically handle meals without the user’s input. By reducing the need for manual bolusing, ABX can improve everyday safety, support better long-term health, and bring us closer to a future where managing T1D is hands-free, accessible, and personalized.