A Mobile App With AI Helps Adults on Insulin Injections Improve Blood Sugar Control

Researchers in a 12-week randomized controlled trial evaluated a decision support system (DSS) designed to help adults with type 1 diabetes using multiple daily insulin injections achieve better blood sugar control. The system combined a mobile application with a Bayesian titration algorithm—a statistical method that learns from data patterns over time.

Eighty-four adults with suboptimal glycemic control (HbA1c of 7.5% or higher) were split into two groups. One group used the DSS alongside Freestyle Libre continuous glucose sensors. The other group received a standard non-adaptive bolus calculator—a tool that calculates insulin doses but doesn't learn or adjust recommendations over time.

The decision support system group achieved a meaningful reduction in HbA1c. Their average HbA1c dropped from 8.6% at the start to 8.1% after 12 weeks. In comparison, the control group's HbA1c fell from 8.6% to 8.5%—a much smaller change. The difference between the two groups was a 0.4% reduction in favor of the DSS, which was statistically significant.

Safety was a priority. The researchers found no cases of severe hypoglycemia (dangerously low blood sugar) or diabetic ketoacidosis (a serious complication) in either group during the trial. This suggests the system's recommendations were safe for participants to follow.

Researchers also tested whether the algorithm's insulin adjustment recommendations aligned with those made by experienced endocrinologists. Thirteen Canadian endocrinologists reviewed historical patient data and made mock insulin adjustments. The researchers then compared these physician recommendations with those the algorithm would have made.

The results showed the algorithm performed comparably to physicians in terms of consistency. When the same physicians reviewed identical data at different times, they only fully agreed with their own previous recommendations about 66% of the time for meal-time insulin and 67% for basal insulin. The algorithm's suggestions showed similar agreement rates with physicians, suggesting automated recommendations can be as reliable as the variable judgments of individual doctors.

For adults struggling with blood sugar control on multiple daily injections, this system offers a new option: automated, weekly guidance tailored to their glucose patterns. Rather than making insulin adjustment decisions alone or relying on occasional clinic visits, people could receive consistent, algorithm-based recommendations through their phone.

While the improvement of 0.4% in HbA1c may sound modest, even small reductions in HbA1c can lower the risk of diabetes-related complications over time. The absence of safety issues in this 12-week trial is also encouraging, though longer follow-up would provide more information about durability and real-world use.

Origin: nature.com (News report). Evidence: News report, corroborated with 3 indexed studies. Type1Cure is an information and intelligence hub, not a medical advice service. This article summarizes published research and does not provide diagnosis, treatment, or personal medical guidance. Always talk to your own care team before changing anything about your Type 1 diabetes management.