Application of data science in risk assessment and early warning

Abstract

The food supply chain has been recognised by the EU as a critical infrastructure, and its complexity is the main cause of vulnerability. Depending on the food matrix, natural and/or deliberate contamination, food‐borne diseases or even food fraud incidents may occur worldwide. Consequently, robust predictive models and/or software tools are needed to support decision‐making and mitigating risks in an efficient and timely manner. In this frame, the fellow participated in data collection and analysis tasks, so as to provide additional predictive models. The working programme, covered a wide range of aspects related to risk assessment including identification of emerging risks (quantitative), microbiological risk assessment, authenticity assessment, spatio‐temporal epidemiological modelling and database formation for hosting predictive microbial models. The training and close integration, in the open‐source, in‐house (German Federal Institute for Risk Assessment (BfR)) developed software tools under the framework of FoodRisk‐Labs (https://foodrisklabs.bfr.bund.de.) for data analysis, predictive microbiology, quantitative microbiological risk assessment and automatic data retrieval purposes allowed for the independent use. Moreover, the fellow actively contributed to the update of the upcoming Yersinia enterocolitica risk assessment, and also in authenticity assessment of edible oils. Over the course of the year, the fellow was closely involved in international and national research projects with experts in the above‐mentioned disciplines. Lastly, he consolidated his acquired knowledge by presenting his scientific work to conferences, and BfR‐internal meetings.

Analysis of the economically motivated food adulteration in China based on 6477 events from 2000 to 2020

“…Although such databases exist, globalization increased the complexity and vulnerability of the food chain, making EMAs more likely to occur. Thus, it is important to establish a more comprehensive and applicable data system to effectively record and report EMA (Pavlidis et al, 2019). In recent years, several FFDs have been utilized for EMA characterization, prediction, and prevention.…”