Human vs. machine in medical search strategy development: a comparative evaluation of ChatGPT-4.1

Abstract

This study examined the potential of generative artificial intelligence, specifically ChatGPT-4.1, to support development of search strategies. Using two Cochrane Review topics as benchmarks, AI-generated MEDLINE strategies were compared to the expert strategies using PRESS (Peer Review of Electronic Search Strategies) criteria. Results show that ChatGPT-4.1 can accurately translate research questions and apply Ovid syntax and Boolean operators, but shows considerable weaknesses in subject heading and text word selection.