Controlled hypoglycemia procedures for Studies 1 and 2 used variable insulin and glucose administration with corresponding euglycemia and hypoglycemia driving sessions (panel A). Venous blood glucose in euglycemia and hypoglycemia for studies 1 and 2 are presented as mean (circles) with standard deviation (whiskers; panel B). In panel C, the setup of car, DMC, and glucose management in both studies is shown in the left panel, and the procedure for creating and evaluating our machine learning models is shown in the right panel. Credit: NEJM AI (2024). DOI: 10.1056/AIoa2300013
Hypoglycemia (hypoglycemia) is one of the most dangerous complications of diabetes and is at high risk during demanding cognitive tasks requiring complex motor skills, such as driving a car. The usefulness of current tools for detecting hypoglycemia is limited by diagnostic delay, invasiveness, low availability, and high costs.
A recent study published in the journal NEJM AI provides a new way to detect hypoglycemia while driving. The research was carried out by scientists from LMU in collaboration with colleagues from the University Hospital Bern (Inselspital), ETH Zurich and the University of St. Gallen.
In their study, the researchers collected data from 30 diabetics while they drove a real car. For each patient, data were recorded once during a normal blood glucose state and once during a hypoglycemic state. To this end, each patient was deliberately placed in a state of hypoglycemia by the medical professionals present in the car. The data collected included driving signals such as car speed and head/gaze movement data, for example eye movement speed.
Subsequently, scientists developed a new machine learning (ML) model that can automatically and reliably detect hypoglycemic episodes using only regularly collected driving data and head/gaze movement data.
“This technology could serve as an early warning system in cars and allow drivers to take necessary precautions before hypoglycemic symptoms impair their ability to drive safely,” says Simon Schallmoser, a doctoral student at the Institute of ‘AI in management of LMU and one of the contributors. researchers.
The newly developed ML model also performed well when only head and gaze movement data were used, which is crucial for future self-driving cars. Professor Stefan Feuerriegel, director of the Institute for AI in Management and project partner, explains: “This study not only shows the potential of AI to improve individual health outcomes, but also its role in improving safety on public roads. »
More information:
Vera Lehmann et al, Machine learning to infer health status using biomedical signals – Detecting hypoglycemia in people with diabetes while driving real cars, NEJM AI (2024). DOI: 10.1056/AIoa2300013
Provided by Ludwig Maximilian University Munich
Quote: AI model provides early warning system for hypoglycemia while driving (February 8, 2024) retrieved February 8, 2024 from
This document is subject to copyright. Apart from fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for information only.
