A Swedish-US Partnership to Develop and Evaluate CAIA-PROM; a Collaborative AI Agent for Item Generation of a Patient-Reported Outcome Measure according to methodological and regulatory standards

Purpose and goal

Patient-reported outcome measures (PROMs) are questionnaires that enable patients to share their health experiences in clinical practice or research. They are central for informing drug development, healthcare/treatment, and policy. In underresourced clinical fields, underserved regions, and low frequency language groups, PROMs are sometimes lacking. Our project aims to address this need with a goal to develop a methodologically and regulatorily robust Collaborative AI-Agent for creating PROMs.

Expected effects and result

Th expected outcome is a Collaborative AI Agent that supports qualitative analysis of patients’ health experiences and that can generate items in Swedish and English according to regulatory standards. The method may as such provide a cost-effective solution and enable broader inclusion of patient groups, countries and linguistic backgrounds in PROM development. The project positions Sweden and the US at the forefront of AI-PROM technologies and establishes long-term collaborations between us.

Planned approach and implementation

We will conduct a proof-of concept case study of a Collaborative AI Agent, built on a Multilingual Large Language Model. The study is conducted in Sweden and the U.S. and focuses on children born with esophageal atresia. Using a rich qualitative dataset, the Collaborative AI Agent´s performance in qualitative analysis will be evaluated for e.g. face validity, inter-rater reliability and feasibility in terms of time required and costs for data analysis compared to human driven analysis.