AI-driven Environmental Protection: The Danish Regions uses AI on soil contamination data

Denmark grapples with concerning soil contamination, posing serious public health risks. The sheer number of potentially affected sites, significant media scrutiny, and budget constraints necessitate an efficient strategy to prioritize site investigations. With the heavy responsibility of monitoring and tackling contamination, the Danish Regions wanted to streamline the process of identifying and prioritizing soil contamination sites.

The regions already had access to a significant amount of data from previous investigations and public databases. Still, it was challenging to manage this volume of data effectively and identify the areas that required immediate action.

They normally tackle this with traditional methods based primarily on static models and existing domain knowledge. However, these approaches were limited as they could not handle the complexity of the data presented or predict soil contamination with sufficient accuracy. This was because the contamination patterns were profoundly complex and could not be easily identified with conventional approaches.

Therefore, the Danish Regions approached Context& and asked for help using their existing data more effectively by applying advanced analytical methods to identify potential contaminated areas and more accurately prioritize their actions.

Context&'s solution involved implementing advanced AI technologies to analyze and process the large amount of data that the regions had access to.

A Proof of Concept (PoC) phase yielded promising results across two critical use cases: establishing the likelihood of site contamination for prioritization and determining the most probable type of contamination on a site. The models' promising accuracies showcased the feasibility and value of our approach in addressing complex environmental challenges.

The two PoCs considered various factors, including geographical data, previous studies, soil consistency, and other relevant parameters. By integrating public and private data sources and expert domain knowledge into the AI models, Context& created a comprehensive solution that could support the regions' decision-making processes.

A key component is prioritizing actions based on the likelihood of soil contamination. By ranking potentially contaminated areas according to their risk level, the regions can optimize their resource allocation and target their efforts to the areas of greatest need. This saves both time and money while increasing the efficiency of their operations.

The Danish Regions had limited time and resources to visit and find the right places where there is contamination. By investing in their own data and utilizing it intelligently, with Context&'s help, they can better assess where to go and take soil samples. This increases the efficiency of their operations and ensures that resources are used where they are most needed for effective environmental protection, saving both time and money.

Based on the success of the two PoCs, Context& and the regions are now embarking on a full-scale project to operationalize these models for broader use. This involves improving the models' performance, developing APIs for communicating results, and integrating the models into an accessible application. In addition, the project is expanding to explore other forms of soil contamination, including PFAS contamination - a recent concern that has dominated the media due to its widespread health impact and which the Danish government has a very strong focus on.

Context&'s solution for the Danish Regions exemplifies how artificial intelligence can drive significant results in organizations with constrained budgets and no tradition of using advanced analytics. This initiative offers a paradigm shift in how the regions and Denmark can respond to the challenges posed by various forms of contamination.