Honored to receive an 
in Philadelphia for our poster presentation on AI-assisted Risk of Bias (RoB) assessment in observational studies.
This recognition reflects the growing importance of applying AI responsibly and transparently in evidence synthesis. Our research explored how large language models can support RoB assessment using a passage-based validation approach, focusing on isolating judgment while maintaining human oversight.
The study evaluated GPT-5 and Gemini across more than 1,000 article-domain assessments using the NavGuide framework for environmental exposure studies. One of the most interesting findings was that while exact agreement with human consensus remains challenging, AI showed strong potential as a second reviewer, especially when supported by transparent evidence passages and adjudication workflows.
Proud of the work by our team and grateful to ISPOR for the recognition.