Data driven food fraud vulnerability assessment using Bayesian Network: Spices supply chain

Recognizing the vulnerabilities that arise in the spices' supply chain for food fraud, determining which products and food fraud types to assess is crucial for ensuring food quality and food safety. In this study, we developed a data driven food fraud vulnerability assessment approach based on a Bayesian Network (BN) and Failure Modes and Effects Analysis (FMEA) to predict the food fraud vulnerability level for products entering Europe including the food fraud types and potential adulterants for each step in the supply chain. The BN model was developed using a dataset based on spice-related fraud cases reported in the European Union (EU) Rapid Alert System for Food and Feed (RASFF) over the period 2005-2020. Three use cases were explored in the study: chilli, black pepper, and turmeric. The model showed a prediction accuracy higher than 95%. The vulnerability factors in the spices' supply chain having the highest prediction accuracy for fraud are closely associated with the product concerned, the site of intervention and the country of origin of the product. A food fraud vulnerability assessment approach developed in this study could support the food industry and authorities to be more efficient in resource allocation for monitoring and verification whilst maximising their opportunity in detecting a fraudulent product.