Exploration of the prediction and generation patterns of heterocyclic aromatic amines in roast beef based on Genetic Algorithm combined with Support Vector Regression

Abstract

Heterocyclic aromatic amines (HAAs) are harmful byproducts in food heating. Therefore, exploring the prediction and generation patterns of HAAs is of great significance. In this study, genetic algorithm (GA) and support vector regression (SVR) are used to establish a prediction model of HAAs based on heating conditions, reveal the influence of heating temperature and time on the precursor and formation of HAAs in roast beef, and study the formation rules of HAAs under different processing conditions. Principal component analysis (PCA) showed that the effect on HAAs generation increases with the increase of heating temperature and time. The GA-SVR model exhibited near-zero absolute errors and regression correlation coefficients (R) close to 1 when predicting HAAs contents. The GA-SVR model can be applied for real-time monitoring of HAAs in grilled beef, providing technical support for controlling hazardous substances and intelligent processing of heat-processed meat products.