The project combines water analysis with data science and creates an innovative tool for water utilities through the advancement of the "K2I trace substance tracker". This cross-laboratory cloud solution enables automated evaluation of non-targeted data and thus helps detecting previously unknown or unnoticed trace substances.
In recent years, numerous developments have taken place in non-targeted screening for the analysis of organic trace substances, which allow the detection of previously unknown or unnoticed substances. In this way, even small changes in a water sample can be detected and unknown pollutants can be identified. Non-targeted analysis (NTA) is a special analytical method in which a chromatographic separation method coupled with high-resolution mass spectrometry (LC-HRMS) is used. In practice, the broad application of the method is hindered by the data volumes of 1 – 2 GB per sample and their difficult and time-consuming evaluation.
In the already completed BMBF project "Artificial and collective intelligence for trace substance tracking in surface water for sustainable drinking water production" (K2I), a demonstrator for a cloud solution was developed in which the laboratories of water suppliers can upload non-target data and automatically evaluate it across laboratories. This networking of existing and newly collected analysis data and meta-data from different laboratories, i.e. a collective intelligence from the water supply and artificial intelligence when processing the data, provides considerable added value.
The results of the BMBF project are now to be further developed as well as validated and optimized in routine use. On one hand, this requires the development of a standardized procedure for the acquisition of even better comparable NTA data. On the other hand, the cloud environment must be expanded and data processing and other IT components optimized to increase data quality and user-friendliness. Further, through exploratory work with raw and drinking water data from participating water suppliers, a larger data pool is created, which enables AI-based applications such as spatially resolved anomaly detection to detect contamination hot spots and the prioritization of new drinking water-relevant substances. Finally, as part of "K2I-Xplore", the demonstrator is to be transferred into an analytical tool that will provide the water suppliers with an innovative tool for NT analysis, which can also be a building block for the risk management required by the Drinking Water Ordinance (TrinkwV).
