AI helps diabetes patients manage glucose levels

Researchers have developed an AI tool that pairs with a smartphone app to automatically guide glucose adjustments in patients with type 1 diabetes.

In a pilot study of 16 patients, the system’s recommendations were deemed safe by expert endocrinologists at a clip of more than 99% over the course of four weeks.

Further, the automated guidance correlated with directions from the experts a respectable 67.9% of the time.

The researchers behind the algorithm work at Oregon Health & Science University in Portland. Nature Metabolism is running their findings in full for free.

In coverage posted by OHSU’s news division, lead study author Nichole Tyler, an MD-PhD candidate, says the team designed the AI using only a mathematical simulator.

“[Y]et, when the algorithm was validated on real-world data from people with type 1 diabetes at OHSU, it generated recommendations that were highly similar to recommendations from endocrinologists,” Tyler notes.

Senior author Peter Jacobs, PhD, a biomedical engineer, adds that other teams have published similar algorithms but OHSU’s effort is the first to test one clinically.

“Very few have shown a statistically relevant outcome,” Jacobs says, “and most do not compare algorithm recommendations with those of a physician.”