Detection of blood glucose and HbA1c from palm perspiration by using artificial neural networks

It has a vital importance to control diabetes disease. Together with matures, disease rates of children is also increasing. Patients have to monitor blood glucose and HbA1c rates regularly. These monitoring needed to be done quick and sterilized. But also pain of patient during these monitoring should be prevented. Shortly, a technique which is not invasive should be used.In this study, it is tried to develop a non-invasive technique and a suitable parameter to the method. Palm perspiration is measured and by calculating various parameters and by using different ANN structures blood glucose and HbA1c rates are tried to be determined. In this study, it is showed how to determine blood glucose and HbA1c rates. As a conclusion; best results are obtained by using difference attributes and 1. monoslope parameters, with Radial Bases artificial Neural Network which indicates an error rate of 24.4% for blood glucose and 14.9% for HbA1c.