Questionnaires on Perceptions of Artificial Intelligence in Health Care Among Health Care Students: Cross-Cultural Translation Into French and Linguistic Validation

Abstract

Background: Artificial intelligence (AI) is rapidly transforming health care by enhancing diagnostic accuracy, optimizing clinical workflows, and supporting decision-making across 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 comparisons. To bridge this gap and enable comparable, evidence-based assessment of AI perceptions among French health care students, rigorous cross-cultural adaptation of validated instruments is essential. Objective: This study aimed to translate, culturally adapt, and linguistically validate 5 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 care training programs. Methods: 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 care students (in medicine, pharmacy, adapted physical activity and health, nursing, and midwifery) to assess clarity, comprehensibility, and acceptability with iterative revisions made until consensus was reached. Results: During forward translation, wording discrepancies were observed for 73.6% (148/201) of expressions, but only 1.0% (2/201) of items required resolution due to meaning differences. In the backward translation step, 97.0% (195/201) of expressions were judged to be conceptually equivalent to the originals; the remaining 3.0% (6/201) of expressions were revised after discussion. Cognitive debriefing with students led to minor wording modifications in 26.4% (53/201) of expressions to improve clarity and readability without altering the underlying concepts. Conclusions: We produced French-language versions of 5 widely used questionnaires assessing health care 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 care students. This work lays the groundwork for subsequent psychometric studies of these French-language versions of questionnaires used in diverse health care training programs.

Ähnliche Arbeiten