Generative artificial intelligence implementation in REDCap

Résumé fourni par la source

Objective: To describe the Research Electronic Data Capture (REDCap) Consortium's initial implementation of generative artificial intelligence (AI) within the REDCap platform using a minimum viable product (MVP) strategy. Materials and methods: Guided by principles of security, optional adoption, and "human-in-the-loop" oversight, we developed and implemented three AI-assisted features: a writing helper, qualitative data summarization, and language translation. Features were disseminated as part of REDCap release 15.0. Results: During the first seven months post-release (January-August 2025), 18 institutions worldwide activated the REDCap generative AI module, with eight reporting sustained use across 1171 projects. At Vanderbilt University Medical Center, 958 projects used at least one feature, generating over 5700 generative AI API calls. Discussion: Early uptake demonstrates feasibility and researcher interest, though adoption depends on local AI tenant infrastructure and governance. Conclusion: The MVP provides generalizable lessons for securely and responsibly deploying generative AI within research electronic data capture systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.