Can AI Help Young Children Start Automated Insulin Delivery?
By April Hopcroft
A short trial is recruiting 30 children between ages 2-5 whose families are interested in trying automated insulin delivery. The study will investigate whether artificial intelligence-based adjustments to insulin pump settings are as effective as without.
Clinical Trials Identifier: NCT06017089
Trial Name: The Pediatric Artificial Pancreas Automated Initialization Trial (PEDAP-AI)
Diabetes Type: Children (ages 2-5) with type 1 diabetes
Trial Sponsor: Dr. Marc Breton at the University of Virginia
What is the trial researching?
This study will examine whether an AI-based assistant can safely and effectively adjust insulin pump settings for young children starting automated insulin delivery (AID).
The trial aims to enroll 30 children with type 1 diabetes who are currently on insulin therapy with multiple daily injections. Participants will use Tandem’s t:slim X2 Control-IQ technology with a Dexcom G6 or G7 continuous glucose monitor (CGM) for eight weeks.
Researchers will measure time in range, time below range (hypoglycemia), and time above range (hyperglycemia).
Why is this trial important?
In recent years, there has been growing recognition that AID systems can help children and caregivers manage type 1 diabetes.
Research suggests the benefits of AID systems extend to young children with type 1 diabetes. Indeed, a 2023 study found that Control-IQ technology led to a 3-hour increase in time in range and 0.5% reduction in A1C among children ages 2-6 with type 1 diabetes.
A similar study found that Medtronic’s MiniMed 780G was safe for use in children ages 2-6, leading to significant improvements in time in range with no diabetic ketoacidosis (DKA) or hypoglycemic events.
Tandem’s Control-IQ is approved for children 6 to 13 and Medtronic’s MiniMed 780G system is approved for ages 7 and up; neither system has been approved for toddlers or young children. To date, the Omnipod 5 is the only AID system cleared for children ages 2 and older.
With advancements in AI, researchers are continuing to explore how this technology could optimize insulin therapy. For instance, a recent study showed that an AI chatbot helped people with type 2 diabetes start on basal insulin and led to better glycemic control.
Typically, starting AID involves many adjustments to insulin pump settings during the first few weeks. In this study, the AI-based assistant may be able to adjust settings such as:
-
Basal rate: This is the rate at which insulin is continuously delivered in small, “background” doses.
-
Insulin-to-carb ratio: This refers to how much rapid-acting insulin is needed per fixed amount of carbohydrates.
-
Insulin sensitivity factor: Also called a correction factor, this is the amount of insulin needed to bring blood glucose into its target range.
-
Target blood glucose: This is based on individual treatment goals – learn more about targets here.
-
Insulin duration: This refers to the length of insulin on board, or how long insulin is active from previous boluses.
By adjusting different pump settings, an AI assistant could reduce the workload for endocrinologists and streamline the process of starting on an AID system for children and their caregivers. If this small pilot study is successful, it could lead to future trials in a larger population of children with type 1 diabetes.
Are you interested?
Your child may be eligible to participate if they:
-
Are 2-5 years old and have had type 1 diabetes for at least three months
-
Weigh at least 20 pounds
-
Take at least five units of insulin per day
-
Are using a Dexcom CGM
-
Have not had a history of DKA in the past six months
-
Have not had a history of hypoglycemia in the past three months
Children who currently use an insulin pump are not eligible.
The trial is currently recruiting in California, Colorado, and Virginia. See a full list of inclusion/exclusion criteria here.
For more information: Contact Dr. Marc Breton at [email protected] or by phone at 434-982-6484.
Learn more about AID systems for children with diabetes:
