Cross-cultural translation and linguistic validation in French of five questionnaires assessing health students’ perceptions of artificial intelligence in healthcare. (Preprint)

Abstract

<sec> <title>BACKGROUND</title> Artificial intelligence (AI) is rapidly transforming healthcare by enhancing diagnostic accuracy, optimizing clinical workflows, and supporting decision‑making across virtually all health disciplines. As AI-driven tools are progressively introduced into health systems, educating future professionals about AI has become a critical priority to ensure safe, ethical, and effective use. Although several validated English-language questionnaires exist to assess medical students’ perceptions and readiness on AI in medicine, no French-language equivalents are currently available, which limits their use in Francophone settings and hampers international comparison. To bridge this gap and enable comparable, evidence‑based assessment of AI perceptions among French health students, rigorous cross‑cultural adaptation of validated instruments is essential. </sec> <sec> <title>OBJECTIVE</title> This study aimed to translate, culturally adapt, and linguistically validate five established English-language questionnaires assessing medical students’ perceptions of AI in medicine, to produce French versions suitable for subsequent psychometric validation and use across health training programs. </sec> <sec> <title>METHODS</title> We followed international guidelines for the cross‑cultural adaptation of self-report measures, combining independent forward translations, reconciliation, backward translation, expert committee review, and cognitive debriefing. Two bilingual translators first produced independent French versions of each questionnaire, which were reconciled into a single draft. A third bilingual translator, blinded to the original instruments, then performed backward translation into English. An expert panel reviewed all versions to ensure conceptual equivalence and to adapt items for applicability across health professions. Finally, cognitive testing was conducted with 38 French health students (medicine, pharmacy, adapted physical activity and health, nursing, and midwifery) to assess clarity, comprehensibility and acceptability with iterative revisions made until consensus was reached. </sec> <sec> <title>RESULTS</title> During forward translation, wording discrepancies were observed for 148 of 201 expressions (73.6%), but only two items (&lt;1%) required resolution due to meaning differences. In the backward translation step, 195 of 201 expressions (97.0%) were judged conceptually equivalent to the originals; the remaining six expressions (3.0%) were revised after discussion. Cognitive debriefing with students led to minor wording modifications in 53 of 201 expressions (26.3%) to improve clarity and readability, without altering the underlying concepts. </sec> <sec> <title>CONCLUSIONS</title> We produced French‑language versions of five widely used questionnaires assessing health students’ perceptions of AI in medicine, following a rigorous cross-cultural translation, adaptation, and linguistic validation process. These instruments preserve conceptual equivalence with their English originals and provide standardized tools to document AI-related knowledge, attitudes, and intentions among French-speaking health students. This work lays the groundwork for subsequent psychometric studies of these French versions in diverse health training programs. </sec>

Ähnliche Arbeiten