Keyur Patel
The health of people depends heavily on the safety of food, and the risk management of food contamination is needed to prevent foodborne diseases. Conventionally, HACCP (Hazard Analysis and Critical Control Points) systems are deployed, but they are restrictive in their ability to control food safety risks, relying on threshold levels and inspection controls. This paper presents a proposal for an AI-based HACCP risk-prediction system that uses machine learning algorithms to forecast the likelihood of food contamination from real-time sensor data. XGBoost, Random Forest, Gradient Boosting, and Logistic Regression machine learning algorithms were deployed on the food contamination dataset from the Kaggle competition to determine the risk of food contamination based on environmental factors such as temperature, humidity, and time spent. XGBoost outperformed all other models, achieving the highest accuracy (90.1%) and ROC AUC (0.92). Various measures were used to assess the accuracy of the XGBoost model's predictions. The performance of the XGBoost model was evaluated using accuracy, precision, recall, F1 score, and ROC curves. This suggests applying predictive analytics to HACCP systems to shift the risk management paradigm from reactive to proactive, enabling food safety managers to address risks before they reach critical thresholds. This is likely to enhance the food safety management system by increasing cost reduction and providing real-time monitoring through the introduction of machine learning into HACCP frameworks. It is believed that this proposed framework will offer significant benefits in predicting food contamination through the HACCP system. To enhance the validity of the models, future studies may use deep learning methods and real-time data processing.