Detection of the blood glucose and haemoglobin A1C with palm perspiration by using artificial neural networks

The invasive measurement techniques that puncture the skin during the detection are generally used for blood glucose and haemoglobin A1C (HbA1C) detection. In this paper, artificial neural network structures were used for the detection of relationship between blood glucose, HbA1C and palm perspiration rate as a non-invasive measurement technique. For this purpose, a comparative study was realized by using feed forward multilayer, Elman and radial basis neural network structures. A data set for 221 volunteers is used for this study. Data of 148 volunteers are used for training of the neural networks and the remaining data were used as test data.