Multivariate Calibration Models to Classify Blood Glucose Levels Non-invasively Based on A New Optical Technique Named Pulse Glucometry

摘要

A novel optical technique for the non-invasive in vivo blood glucose concentration (BGL) measurement, named "Pulse glucometry", was combined with two discriminant analyses using Artificial Neural Network (ANN) and Support Vector Machines Classification (SVMs). The total transmitted radiation intensity (I~λ) and the cardiac-related pulsatile changes superimposed on I~λ in human adult fingertips were measured over the wavelength range from 900 to 1700 nm using a very fast spectrophotometer with sampling speed of 100 spectra/s, obtaining a differential optical density (AOD~λ) related to the blood component in the finger tissues. Subsequently, two discriminant analyses also attempted to distinguish BGL levels according to such two criterions of diabetic screening as (1) a case of fasting blood sugar and (2) a case of 2-hour postprandial blood sugar. Using 183 paired data sets, in which measured BGL values ranged from 89.0-219 mg/dl (4.94-12.2 mmol/1), we found that the two discriminant analyses also showed over 80% accuracy for the case (1) and over 65% accuracy for the case (2). These results provide a preliminary evidence that Pulse glucometry with a discriminant type calibration using ANN and SVMs appears useful and promising for the screening of BGL levels.