The official monitoring and testing of food products, such as meat and meat products, for suitability for consumption currently involves time-consuming laboratory analysis. This makes it impossible to continuously monitor the microbiological and biochemical changes in products between production and consumption. However, innovative technologies and artificial intelligence (AI) offer new ways of predicting food quality and safety continuously. Such an approach was developed within the consortium project “AI-augmented approaches to assess food shelf life 2030” (ZL2030), in which the MRI participated.
AI-augmented approaches to assess food shelf life 2030
The ZL2030 project aimed to develop scenarios for enhancing food monitoring and consumer protection across the entire food supply chain by employing innovative analytical techniques and AI. To this end, data related to quality and safety of foodstuffs were collected. This dataset was then expanded using data from non-targeted analytical methods to form the basis for developing a “digital twin” of the foodstuff in question. The feasibility of this approach was tested using minced pork as a model.
The MRI's Institute of Safety and Quality of Meat was one of nine institutions involved in this consortium project. The MRI was responsible for producing the samples, conducting storage trials at its pilot plant and transporting the samples to project partners while maintaining the correct temperature. The MRI also performed laboratory analyses and sensory evaluations of the minced meat.
Within the project, more than 7,000 samples from 31 trials were examined. The resulting database contains microbiological, physical, and biochemical data on parameters such as bacterial counts and species identification, pH value, L*a*b* colour and the extent of fat oxidation. Innovative analytical techniques such as metabolome and volatilome analyses, spectroscopy, and next-generation sequencing were also employed. All datasets from each of the project partners were compiled in an online database (openBIS ).
Data obtained during the project can now be used to create a “digital twin” of minced pork. When combined with innovative supply chain monitoring technologies, the rapid techniques developed during the project can instantly provide the “digital twin” with data from current measurements taken during production, distribution or sale. This enables the “digital twin” to predict the food's actual expiry date in real time based on current data.
Against food waste
More accurate shelf life predictions could help reduce future food waste. Throughout the project, the Research Centre for German and European Food Law at the University of Bayreuth will examine issues related to food law. This includes questions relating to labelling on packaging, such as “best before” and “use by” dates. Ultimately, the Research Centre will develop recommendations for an appropriate course of action based on AI-driven predictions of the food's actual condition.
Project funding
With support from Federal Ministry of Agriculture, Food an Regional Identity by decision of the German Bundestag
