Noninvasive glucometer model using partial least square regression technique for human blood matrix

Abstract

In this article, we have highlighted the partial least square regression (PLSR) model to predict the glucose level in human blood by considering only five variants. The PLSR model is experimentally validated for the 13 templates samples. The root mean square error analysis of design model and experimental sample is found to be satisfactory with the values of 3.459 and 5.543, respectively. In PLSR templates design is a critical issue for the number of variants participating in the model. Ensemble consisting of five major variants is simulated to replicate the signatures of these constituents in the human blood, i.e., alanine, urea, lactate, glucose, and ascorbate. Multivariate system using PLSR plays important role in understanding chemometrics of such ensemble. The resultant spectra of all these constituents are generated to create templates for the PLSR model. This model has potential scope in designing a near-infrared spectroscopy based noninvasive glucometer.

Comparison of Linear and Non-Linear Regression Models for Non-Invasive Blood Glucose Measurement

“…Section 1 presents the best results in the literature. Our experimental results prove that our chosen technique as well as the regression model are better than the other methods (Parab et al, 2010;Uwadaira et al, 2015;Mougiakakou et al, 2006;Perez-Gandia et al, 2010;Robertson et al, 2011;Pappada et al, 2011;Zecchin et al, 2012;Ming and Raveendran, 2009;Zhua et al, 2019;Wu et al, 2017).…”

“…These approaches have reduced the Standard Error (SE) from 40 mg/dL using unfiltered module to 15.63 mg/dL using a Chebyshev filter. Parab et al (2010) used the Partial Least Squares (PLS) regression model to design a noninvasive glucometer. The experimental results were satisfying, with values of 3.459 for the RMSE.…”

Multivariate calibration of NIR spectroscopic sensors for continuous glucose monitoring

“…It should be noted that some studies were based on simulated data [102][103][104]. However, the results obtained for the models built in these studies were limited to those simulated data so we do not discuss them in this review.…”

Prospects and limitations of non-invasive blood glucose monitoring using near-infrared spectroscopy

“…For in vitro experiments various chemical components/blood constituents in the human blood, i.e. protein, lactate, glucose, lipids and amino acids can be mixed with glucose to measure the effect of other composites of blood [106,107]. In case of in vitro experiments, the NIRS shows positive results and experimental conditions are less complicated as compared to in vivo measurements [108].…”