Non- invasive diabetes detection and classification using breath analysis

Diabetes is a major problem affecting millions of people today and if left unchecked can create enormous implication on the health of the population. Among the various non invasive methods of detection, breath analysis presents an easier, more accurate and viable method in providing comprehensive clinical care for the disease. This paper examines the concentration of acetone levels in breath for monitoring blood glucose levels and thus predicting diabetes. The analysis uses the support vector mechanism to classify the response to healthy and diabetic samples. For the analysis ten subject samples of acetone levels are taken into consideration and are classified according to three labels which are healthy, type 1 diabetic and type 2 diabetic.