Improving AI Readiness Among Medical Trainees Through an Interdisciplinary Health Datathon: Mixed Methods Evaluation (Preprint)
Authors:
- Emily LeventhalELEmily Leventhal
- Sahil SureshSSSahil Suresh
- Arvind RajanARArvind Rajan
- Kaden BunchKBKaden Bunch
- Bhavana KunisettyBKBhavana Kunisetty
- Maya V. RoytmanMRMaya V. Roytman
- Jennifer IpeJIJennifer Ipe
- Samantha MallahanSMSamantha Mallahan
- Michael YaoMYMichael Yao
Preprint
- 3 July 2026
- preprint
- Published by JMIR Publications Inc. in JMIR Preprints
Abstract
BACKGROUND Artificial intelligence (AI) is transforming healthcare, yet medical trainees often lack formal AI education. Further, clinician resistance is a major barrier to AI integration into practice. Experiential learning approaches such as datathons may offer a scalable pathway to improve AI literacy and acceptance among medical trainees. OBJECTIVE To evaluate whether participation in a virtual, team-based health AI datathon improves medical AI readiness, technology acceptance, and digital health literacy among trainees. METHODS We organized a 3-week virtual datathon themed “Empowering Patients Through AI” through a national trainee-led nonprofit with both academic and industry partners. Teams of trainees analyzed one of four provided public healthcare datasets to develop patient-centered AI solutions. Project submissions were analyzed using reflexive thematic analysis. Pre- and post-event surveys assessed AI readiness, technology acceptance, and digital health literacy using adapted Medical Artificial Intelligence Readiness Scale for Medical Students (MAIRS-MS) and Technology Acceptance Model (TAM) scales. Within-participant differences in composite scores were evaluated using Wilcoxon signed-rank tests. Free-text reflections were analyzed using data-driven consensus clustering of reviewer-extracted codes. RESULTS The program attracted registrants (n=292) from 75 institutions, with 31 teams (n=149 participants) submitting final projects across 10 clinical domains. Six patient empowerment themes were identified among submitted projects, with symptom forecasting (39%), self-management (32%), and health literacy (26%) being most common. Among paired respondents (n=69), medical AI readiness (mean 3.61 to 4.24), AI technology acceptance (3.93 to 4.36), and digital health literacy (3.67 to 4.23) all increased significantly (all P<.001). The largest gains were in identifying AI-appropriate problems, explaining core AI concepts, and using digital tools for self-management. Qualitative analysis of free-text reflections identified seven themes, with teamwork, interdisciplinary collaboration, and motivation appearing most frequently, all associated with strongly positive sentiment expressed by participants. CONCLUSIONS Participation in a health AI datathon improved AI readiness and acceptance while generating diverse, patient-centered solutions. Datathons represent a scalable experiential learning model to prepare trainees for effective engagement with AI in healthcare, while creating opportunities for trainees to explore and develop novel health technology solutions.Keywords
This publication has 37 references indexed in Scilit:
- Using thematic analysis in qualitative researchJournal of Medicine, Surgery, and Public Health, 2025
- Innovating Challenges and Experiences in Emory Health AI Bias Datathon: Experience ReportJournal of Imaging Informatics in Medicine, 2025
- A population-based investigation of participation rate and self-selection bias in momentary data capture and survey studiesCurrent Psychology, 2023
- A practical guide for conducting qualitative research in medical education: Part 2—Coding and thematic analysisAEM Education and Training, 2021
- Healthcare Hackathons Provide Educational and Innovation Opportunities: A Case Study and Best Practice RecommendationsJournal of Medical Systems, 2016
- Ward’s Hierarchical Agglomerative Clustering Method: Which Algorithms Implement Ward’s Criterion?Journal of Classification, 2014
- Algorithms for hierarchical clustering: an overviewWIREs Data Mining and Knowledge Discovery, 2011
- K-Means Clustering2011
- Term-weighting approaches in automatic text retrievalInformation Processing & Management, 1988
- Silhouettes: A graphical aid to the interpretation and validation of cluster analysisJournal of Computational and Applied Mathematics, 1987