LEVERAGING MACHINE LEARNING TECHNIQUES FOR THE PREDICTION AND ENHANCEMENT OF FOOD SAFETY STANDARDS IN NIGERIA: A DATA-DRIVEN APPROACH TO IDENTIFYING AND MITIGATING CONTAMINATION RISKS

Abstract

Food safety in Nigeria is a critical public health concern, with an estimated 200,000 annual foodborne illness cases straining a system reliant on traditional methods like inspections and laboratory testing. These approaches, hindered by high costs, limited scalability, and delays, struggle to address the growing complexity of contamination risks in the country’s decentralized food supply chain. This study investigates the use of machine learning (ML) to predict food safety risks, leveraging data from 2015 to 2023, including over 50,000 data points from government reports, inspection records, and public datasets. Focusing on microbial contamination, chemical residues, and illness reports, the research tested three ML models: Random Forest (RF) with 89% accuracy, Support Vector Machines (SVM) with 85% accuracy, and Neural Networks (NN) with the highest performance at 91% accuracy, 89% precision, 88% recall, and an F1-score of 88%. The NN model excelled in predicting risks tied to fresh produce and processed foods, which account for about 60% of foodborne illnesses in Nigeria. Findings suggest ML, particularly NN, could reduce illness rates by up to 20% if scaled. The study highlights ML’s potential to revolutionize food safety protocols, offering more accurate and reliable predictions than traditional methods. However, challenges such as poor data quality and availability could limit effectiveness. Addressing these barriers could enhance food safety management in Nigeria, improving public health and providing a scalable solution to a pressing national issue.