JMIR Publications

ABSTRACT

Medical schools worldwide are integrating artificial intelligence into their curricula, but this effort addresses the top of the competency stack while ignoring its foundation. A meaningful capacity to work with AI tools requires prior mastery of data handling, structuring, and sharing practices that are rarely taught systematically. This paper argues that health data workflow literacy — the practical ability to manage, structure, and share data in reproducible and interoperable ways — should be recognised as a core interprofessional competency in health professions education, distinct from and prerequisite to AI literacy. I define health data workflow literacy and its component competencies; analyse the gap in current frameworks (IPEC, WHO, AAMC); and illustrate how these competencies can be taught using free, open-source tools, drawing on teaching experience at the Faculty of Medicine of the University of Porto. Current competency frameworks address clinical informatics and AI literacy but omit the intermediate layer of data workflow literacy. This gap leads to graduates who can discuss AI conceptually but cannot structure a dataset, manage references as machine-readable metadata, track changes to an analysis, or produce a reproducible report. Free tools such as Zotero, Git, and Quarto can teach these competencies within existing courses without requiring dedicated informatics tracks. Health data workflow literacy is a prerequisite for meaningful engagement with AI tools, clinical information systems, and interprofessional collaboration. It should be embedded longitudinally in health professions curricula and explicitly included in major competency frameworks. The tools to address this gap are available and free; what is needed is curricular and institutional will.