Food Safety - Supplier Sampling Plan

In food supply chains, Food Business Operators (FBO’s) need to develop a monitoring plan to manage food safety risks. This project aims to develop a tool that can be used by food business operators (FBO’s) and auditors to conduct risk-based monitoring. For this purpose, an interactive tool is to be developed where users can fill in a list of suppliers and products and get risk ranking as output, using a large, recently published database of analytical results gathered by National Food Safety Authorities. In the project an algorithm is to be developed that is capable of deriving which food safety risks are most likely to occur building on a (public) database of analytical results, derived from over 15 million samples taken by EU food safety authorities (currently available on Github / Zenodo).

The MVP takes in a list of ingredients per end-product (the product recipe), quantities of the ingredient used and countries where these ingredients originate from (additional characteristics optional). The algorithm ranks chemical food safety risks per ingredient based on product type, country of origin, trends in underlying data and seasonality (optional), the risk ranking should weigh in the amount of raw material used but also the severity of the risk. The MVP also takes in FBO’s characteristics, such as (as a minimum) the (EU-) country where it is based and (optional) production process characteristics. The MVP has a UI where a user can upload or fill in data, generate a risk ranking / monitoring plan. The output should be explainable, e.g. the UX should allow the user to gain insight in what data has been used to derive a certain ranking and the user should be able to export the ranking as well as the underlying statistics / calculations / reasoning.

The fully completed version of the project is capable of identifying food safety risks and generating a full monitoring plan (which samples to take, where and when) based on not only product and supplier data, but also on production process characteristics and is capable of providing a monitoring plan not only for chemical risks but also to control microbiological and physical risks.