AI tools used in EPA’s Systematic Review Process

Artificial intelligence (AI) has the potential to dramatically reduce the time required to identify and aggregate potentially relevant data for use in chemical assessments. Currently, EPA uses software with AI methods to more efficiently complete resource-intensive tasks like screening literature for relevance and data extraction. The presentation summarizes the software and AI methods currently used in chemical assessments and investigations into how generative AI (GenAI) could be used for even more tasks. 

Impact/Purpose

EPA currently uses artificial intelligence (AI) methods to reduce the resources required to complete chemical assessments. With the increased availability and performance of generative AI (GenAI), we are investigating how GenAI can be utilized for better performance compared to traditional machine-learning (ML) methods and for more tasks that haven't been able to be addressed by ML methods before (e.g., study evaluation). Continued investigation and implementation of AI into tools for wider adoption in chemical assessment workflows could dramatically reduce the time required to identify and aggregate potentially relevant data for use in chemical assessments reducing the overall time required to complete an assessment.

Citation

Watford, S. AI tools used in EPA’s Systematic Review Process. Federal-State Toxicology Risk Analysis Committee (FSTRAC), Washington, DC, August 12, 2025.