The percentage of people who had diabetes in 2019 in the U.S. (11.3% or 37.3 million) and the world (9.3% or 463 million) continue to creep higher. Measuring blood glucose levels is part and parcel of managing diabetes. The worldwide demand for help for people with diabetes drives massive technology development efforts for wearable, carryable, and implanted digital glucose monitoring devices. Researchers at the George Mason University Electrical and Computer Engineering Department and other research institutions recently published a report in IEEE Xplore that describes their progress in developing an electronic nose (E-Nose) system that can detect and precisely quantify blood glucose levels.

The George Mason team developed an E-Nose system with a metal oxide gas sensor array. Twelve sensors collect 14 gases from exhaled breath. In the published study, the researchers collected exhaled breath from 41 subjects. The next step entailed machine learning classification and regression models to detect and measure blood glucose levels based on the detected gases. According to the study report, the E-Nose system identified blood glucose levels with 90.4% accuracy.

The George Mason researchers’ enabling technology may eventually see life in clinical or consumer devices. At this point the team focuses on increasing measurement precision and assembling additional datasets to use for machine-learning to build even more effective algorithms. According to the study report, the George Mason team believes the study supports the machine-learning-enabled E-Nose system as an “efficient and precise method to achieve low-cost and non-invasive disease diagnosis.” If diabetics could monitor their blood sugar levels simply by blowing into a tube instead of getting blood samples from a finger prick, that could be a major improvement in how we manage the chronic condition.