AI Steps Up in Diabetes Care: Outperforming Doctors in Key Areas

JUST IN: diabetes care is getting a shake-up. A new AI tool, built to interpret continuous glucose monitoring (CGM) data, outperforms doctors in key areas like empathy and actionability. But let's not get carried away, these systems aren't ready to make standalone therapeutic decisions.

LLMs in the Diabetes Arena

Between October 2025 and February 2026, researchers put a retrieval-grounded, large language model-based conversational agent (CA) to the test. The goal? To see if it could help patients better understand their CGM data and prepare for diabetes consultations. The results? Pretty wild.

They crafted 12 CGM-informed cases from publicly available data, then had six senior UK diabetes clinicians review and respond to them. The experiment was rigorous, each case received input from clinicians and the AI, with responses rated on six quality dimensions by independent clinicians.

A Surprising Outcome

to the numbers. The AI tool garnered a mean quality score of 4.37, significantly higher than the clinicians' 3.58. That's a difference of 0.782 points, and it's not trivial. Where the AI really shined was in empathy, scoring 1.062 points higher, and in actionability, with a lead of 0.992 points.

But here's the kicker: safety flags, those red flags that make you pause and think, 'Could this be risky?' were equally rare in responses from both AI and humans. Just 0.7% of the time, to be exact.

Not Ready to Fly Solo

So, does this mean AI is ready to replace your diabetes doc? Not quite. While the CA shows promise as a supplementary tool, it can't yet make autonomous therapeutic decisions or go unsupervised in real-world scenarios. The labs are scrambling to refine this technology, but for now, it's more of a co-pilot than a pilot.

With AI showing strong empathy and actionable insights, one has to wonder: Is the future of healthcare a hybrid of human and machine? Only time, and more tests, will tell. But one thing's certain, the leaderboard in diabetes care is shifting.