Bitterness intensity prediction of berberine hydrochloride using an electronic tongue and a GA-BP neural network

Abstract

The aim of this study was to predict the bitterness intensity of a drug using an electronic tongue (e-tongue). The model drug of berberine hydrochloride was used to establish a bitterness prediction model (BPM), based on the taste evaluation of bitterness intensity by a taste panel, the data provided by the e-tongue and a genetic algorithm-back-propagation neural network (GA-BP) modeling method. The modeling characteristics of the GA-BP were compared with those of multiple linear regression, partial least square regression and BP methods. The determination coefficient of the BPM was 0.99965±0.00004, the root mean square error of cross-validation was 0.1398±0.0488 and the correlation coefficient of the cross-validation between the true and predicted values was 0.9959±0.0027. The model is superior to the other three models based on these indicators. In conclusion, the model established in this study has a high fitting degree and may be used for the bitterness prediction modeling of berberine hydrochloride of different concentrations. The model also provides a reference for the generation of BPMs of other drugs. Additionally, the algorithm of the study is able to conduct a rapid and accurate quantitative analysis of the data provided by the e-tongue.

Keywords: berberine hydrochloride; bitterness intensity; bitterness prediction model; electronic tongue; genetic algorithm-back-propagation neural network.

Figure 1

Flow chart of the GA-BP. GA-BP, genetic algorithm-back-propagation neural network.

Figure 2

Concentration and corresponding rank bitterness intensity of the nine samples (S1–S9).

Figure 3

(A) Fitness curve of the GA and (B) training curve of the GA-BP. GA-BP, genetic algorithm-back-propagation neural network.

Figure 4

Fitness curve of the GA (cross-validation). GA, genetic algorithm.

Figure 5

Prediction values vs. the I₀ of the GA-BP (cross-validation). I₀, rank bitterness intensity; GA-BP, genetic algorithm-back-propagation neural network.

Figure 6

(A) Prediction values of I₀ vs. the I₀ of the GA-BP and (B) standardized residuals of the GA-BP model. I₀, rank bitterness intensity; GA-BP, genetic algorithm-back-propagation neural network.

Figure 7

Prediction values of I₀ vs. the I₀ of MLR, PLS regression and BPNN models and optimization of the number of latent variables in the PLS regression model. Prediction values for (A and B) MLR and (C and D) PLS regression. (E and F) Optimization of the number of latent variables in PLS regression. (G and H) Preduction values for the BPNN. I₀, rank bitterness intensity; MLR, multiple linear regression; PLS, partial least squares; BPNN, back-propagation neural network; R, correlation coefficient; R², determination coefficient; RMSECV, root mean square error of cross-validation.

Figure 8

R values (cross-validation) of the four methods. R, correlation coefficient; MLR, multiple linear regression; PLSR, partial least squares regression; BPNN, back propagation neural network; GA-BP, genetic algorithm-back-propagation neural network.