A novel breath analysis system for diabetes diagnosis

A noninvasive breath analysis system for diabetes diagnosis is proposed in this paper. It utilizes commercial chemical sensors to detect acetone in human breath. The device is portable, fast and easy to operate. Experiments with real breath samples from both inpatient and outpatient diabetics validated the accuracy of this system. An optimal sensor array for diabetes diagnosis is decided according to the classification accuracy. We hope this system could contribute a step toward a practical system that could be used for diabetes screening.