Development of Robust Calibration Models Using Support Vector Machines for Spectroscopic Monitoring of Blood Glucose

• Published in: Analytical chemistry (Washington) Vol. 82; no. 23; pp. 9719 - 9726
• Main Authors: Barman, Ishan, Kong, Chae-Ryon, Dingari, Narahara Chari, Dasari, Ramachandra R, Feld, Michael S
• Format: Journal Article
• Language: English
• Published: Washington, DC American Chemical Society 01.12.2010
• Subjects: Accuracy; Analytical chemistry; Artificial Intelligence; Blood; Blood Glucose - analysis; Calibration; Chemistry; Exact sciences and technology; Fluctuations; Glucose; Heterogeneity; Human populations; Humans; Least-Squares Analysis; Scattering; Spectrometric and optical methods; Spectrum analysis; Spectrum Analysis, Raman - methods; Spectrum Analysis, Raman - standards; Turbidity; Glucose; Support vector machine; Mixture; Blood; Heterogeneity; Tissue; Improvement; Accuracy; Absorption; Dynamics; Raman spectrometry; Vector; Monitoring; Human; Linear regression; Sample; Prediction; Concentration; Calibration; PLS regression; Regression model; Morphology; Transfer; Parameter; Skin; Technique; Detection; Application
• ISSN: 0003-2700; 1520-6882; 1520-6882
• DOI: 10.1021/ac101754n

Sample-to-sample variability has proven to be a major challenge in achieving calibration transfer in quantitative biological Raman spectroscopy. Multiple morphological and optical parameters, such as tissue absorption and scattering, physiological glucose dynamics and skin heterogeneity, vary significantly in a human population introducing nonanalyte specific features into the calibration model. In this paper, we show that fluctuations of such parameters in human subjects introduce curved (nonlinear) effects in the relationship between the concentrations of the analyte of interest and the mixture Raman spectra. To account for these curved effects, we propose the use of support vector machines (SVM) as a nonlinear regression method over conventional linear regression techniques such as partial least-squares (PLS). Using transcutaneous blood glucose detection as an example, we demonstrate that application of SVM enables a significant improvement (at least 30%) in cross-validation accuracy over PLS when measurements from multiple human volunteers are employed in the calibration set. Furthermore, using physical tissue models with randomized analyte concentrations and varying turbidities, we show that the fluctuations in turbidity alone causes curved effects which can only be adequately modeled using nonlinear regression techniques. The enhanced levels of accuracy obtained with the SVM based calibration models opens up avenues for prospective prediction in humans and thus for clinical translation of the technology.