Noninvasive Glucose Detection Using Joint Methods of Scattering-Photo-acoustic Spectroscopy

Abstract

Diabetes is a worldwide spreading disease according to the World Health Organization (2009). In order for diabetic patients to adhere a strict diet, exercise and medication; an accurate and reliable ways glucose measurement must be available. The frequent puncturing of skin in the invasive measuring devices may cause an increase in skin permeability; at the same time noninvasive techniques produce a high percentage of error compared to the invasive techniques. A Joint noninvasive technique was proved to reduce the root mean squared error of prediction. In this study two noninvasive techniques were used in the measurements; the scattering spectroscopy technique based on the Raleigh scattering theory and the linear relation between glucose concentration and the scattering angle, and the Photo-Acoustic Spectroscopy technique. The equations relating each of the methods with glucose concentration was derived. The reading of each method was evaluated against a reference invasive device (Omnitest® plus from B|Braun); using the Clarke Error Grid Analysis (EGA) and the Mean Squared Error of prediction (MSE). The Linear equation relating glucose concentration with the scattering angle produced the best measurements among the three with 63.4% of the reading falling into region A, 36.6% into region B and MSE of (525.8). The three measurements were then combined using the measurements of central tendency; the mean, median and mode. The mean results were the best since 68.2% of the result fell into region A, 31.8% into region B and the MSE was reduced to (452.2).