Development of Milk Quality Index and Its Prediction Using Regression and Artificial Neural Network Approach

ABSTRACT Milk is one of the nutritious foods consumed regularly by people. It is highly perishable and prone to spoilage quickly after milking from animals. The study aimed to develop a Milk Quality Index (MQI) comprising physicochemical, microbiological, and volatile parameters and to create a predictive model based on this index for evaluating milk quality. In this study, milk was stored for 0–240 min, and its physicochemical properties (fat, SNF, protein, lactose, density, color, and pH), microbial analysis (MBRT and TPC), and volatile compounds were assessed using an E‐nose at 30‐min intervals. The major volatile compounds responsible for spoilage were identified using the Partial Least Squares regression—Variable Importance in Projection method. Furthermore, Principal Component Analysis was performed on physicochemical, microbial, and volatile compounds obtained from the VIP score to group the milk samples based on spoilage. The MQI was developed for grouped milk samples and found that an MQI ≤ 3.0 indicates a fresh sample, an MQI between 3 and 6 indicates the beginning of spoilage and requires heat treatment for consumption, and an MQI ≥ 6 indicates spoiled. Multiple linear regression and Artificial Neural Network models were developed by relating physicochemical and microbial analysis to MQI to predict milk quality. This study provides insight into the development of MQI models for real‐time milk spoilage prediction and decision‐making in the dairy industry.