Comparative studies for multivariate model of human blood tissue using gaussian and lorentz oscillators and analysis using PLS regression

Abstract

The paper presents the modeling of the Multivariate model of the Human blood tissue using Gaussian and Lorentz Oscillators. The authors have demonstrated the spectroscopic modeling of the tissue in the region of near infrared i.e 2000nm to 2500 nm, using the concept of fundamental oscillators responsible for the energy of the composition under study. Usually the two approaches i.e. Gaussian and Lorentz are very popular of which later is complex for modeling the system as it required additional parameter i.e. number of oscillators, Line width of Oscillators and Strength of each oscillator. The general model is customized for the human tissue consisting of five major constituents i.e. serum Alanine, serum urea, HDL Lactate and Ascorbate and Glucose in the physiological concentration in the human tissue. Here the ensemble of around 1000 spectra’s were generated for the template in the physiological concentration of normal population. The Partial Least Square Regression studies were performed for the both the approached for validation model with RMSE and Clark Error Grid analysis.