New AI system tracks blood sugar levels with noninvasive ECG data

For decades, noninvasive blood glucose monitors have been one of the holy grails in the field of medical diagnostic equipment research and development,media reported. From watches to contact lenses, none of them can continuously monitor blood sugar levels without harming the skin, despite the emergence of a variety of seemingly promising innovations. Now, researchers from the University of Warwick are showcasing their latest forward-looking technology and using artificial intelligence (AI) to detect hypoglycemia events from simple cardiosignal signals.

New AI system tracks blood sugar levels with noninvasive ECG data

Leandro Pecchia, author of the new study, points out that finger blood collection is never pleasant and in some cases particularly difficult, such as at night or in the face of childhood patients, and that their latest technology has been innovative in using artificial intelligence to automatically detect hypoglycemia by beating several ECG slots. And ECG can be detected in any case including sleep.

A key breakthrough for the University of Warwick team was the development of an AI system that could learn the ECG rhythms of individual patients.

As shown in the figure below, the ECG measurements between the two subjects that sent a low blood sugar signal may be completely different, meaning that the only way out is to develop an AI system to detect individual fluctuations in each patient.

In response, Pecchia said: “These striking differences may explain the failure of previous studies that have used ECG to detect hypoglycemia events. “

In healthy volunteer tests, the system detected hypoglycemia events with 82 percent accuracy, the researchers said.

Of course, this is not the first noninvasive blood glucose monitoring system to be proven effective in the early experimental phase. Next, researchers need to do a lot of work to validate and refine the use of this technology in a larger patient population.