Closing the data gaps: systematic identification of hazardous chemicals using AI

Research Project, 2025 – 2029

Chemical regulation depends on risk assessments that require detailed data on hazardous chemicals. Currently, this data is primarily generated through resource-intensive and costly in vivo exposure experiments, which can be performed for a small fraction of the chemicals in use. New approach methods, particularly computational QSAR models, offer alternatives but have not yet reached the accuracy needed to replace experimental testing on a larger scale.In this project, we will address the critical lack of hazard data that limits effective risk assessment. We will conduct a comprehensive and systematic assessment of the hazardous properties of the hundreds of thousands of chemicals in use today, identifying previously overlooked risks. To ensure high reliability, we will develop advanced AI methods capable of predicting hazardous properties – including toxicity, persistence, and bioaccumulation – directly from chemical structures. Preliminary data already show that our approach significantly outperforms conventional QSAR methodologies. We will also establish an early warning system that monitors databases, patents, and scientific literature for emerging, potentially harmful chemicals.By closing essential data gaps, this project will improve the effectiveness and precision of chemical regulation. Our outcomes will, thus, support green chemistry initiatives, chemical substitution, and overall sustainable use of chemicals.