Multivariate Regression and Discriminant Calibration Models for a Novel Optical Non-Invasive Blood Glucose Measurement Method Named Pulse Glucometry

摘要

A novel optical non-invasive in vivo blood glucose concentration (BGL) measurement technique, named "Pulse Glucometry", was combined with a kernel method; support vector machines. 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, obtaining a differential optical density (ΔOD~λ) related to the blood component in the finger tissues. Subsequently, a calibration model using paired data of a family of ΔOD~λs and the corresponding known BGLs was constructed with support vector machines (SVMs) regression instead of using calibration by a conventional primary component regression (PCR) and partial least squares regression (PLS). Secondly, SVM method was applied to make a nonlinear discriminant calibration model for "Pulse glucometry." Our results show that the regression calibration model based on the support vector machines can provide a good regression for the 101 paired data, in which the BGLs ranged from 89.0-219 mg/dl (4.94-12.2 mmol/I). The resultant regression was evaluated by the Clarke error grid analysis and all data points fell within the clinically acceptable regions (region A: 93%, region B: 7%). The discriminant calibration model using SVMs also provided a good result for classification (accuracy rate 84% in the best case).