Physicians Prefer OpenEvidence Over Other AI Tools in Stanford-Harvard Study

OpenEvidence said in a press release that physicians selected its clinical AI platform more often than any other chatbot in the independent NOHARM study, conducted by researchers from Stanford University and Harvard University. The study compared 45 large language models and 4 clinical AI systems across more than 4,000 possible clinical actions.

In the randomized evaluation of 101 board-certified physicians in the United States, participants who could choose any AI resource turned to OpenEvidence in 22.3% of their responses. This exceeded the combined 19.8% usage for ChatGPT, Claude, Gemini, and all other external AI platforms.

OpenEvidence noted that physicians who participated in the study made their choices without compensation or incentives. The company stated that its responses are sourced and cited from peer-reviewed medical literature.

The press release also highlighted OpenEvidence’s new EvidenceGrade feature, which grades and visualizes the quality of the medical evidence used in each AI-generated answer. The feature applies the GRADE framework used by organizations such as the World Health Organization and Cochrane.