The Context-Feedback-Visibility (CFV) framework: mechanisms of AI-Empowered clinical decision-making training in pharmacology education

Pharmacology education wrestles with a stubborn challenge. Students can recite the mechanisms and indications of countless drugs, yet when they stand before a real patient, that declarative knowledge often fails to crystallize into a confident medication decision. The gap signals something deeper than insufficient practice. It points to a structural misalignment between two competing logics: the disciplinary logic that organizes curricula around neat drug categories, and the work logic that demands knowledge be mobilized flexibly around patient problems. This study pursued two interconnected aims. The first was to empirically test whether an AI-empowered work logic restructuring model could strengthen clinical decision-making competence in pharmacology education. The second, deeper aim was to uncover the mechanisms that explain why and how coupling AI with curriculum restructuring produces its effects. A four-group quasi-experimental design was implemented over two consecutive semesters at a higher vocational pharmacy college in China. The Full Intervention group ( n = 42) experienced a curriculum rebuilt around typical medication tasks and supported by a full suite of AI tools. The AI-Only group ( n = 40) used identical AI tools within a conventional drug-classification-based syllabus. The Curriculum-Only group ( n = 41) followed the restructured curriculum without AI support. The Traditional Instruction group ( n = 41) received standard case-based lectures. We assessed knowledge application through a standardized case analysis examination, class engagement via the Utrecht Work Engagement Scale for Students, and learning experience with a custom satisfaction questionnaire. The Full Intervention group outperformed every other group across all outcome measures (all p < 0.001, η²p 0.17–0.26). Its mean knowledge application score reached 82.5 (SD = 6.3), compared with 70.2 (SD = 7.8) in the AI-Only group, 74.3 (SD = 7.1) in the Curriculum-Only group, and 65.8 (SD = 8.4) in the Traditional Instruction group. Crucially, the four groups performed comparably on factual recall items ( p = 0.796), confirming that the intervention enhanced knowledge mobilization rather than mere retention. The Full Intervention group also posted markedly higher class engagement (M = 5.92 vs. 4.69, p < 0.001) and learning satisfaction (M = 4.51 vs. 3.52, p < 0.001). From these empirical patterns we distill the Context-Feedback-Visibility (CFV) Framework, which identifies three sequential mechanisms through which AI coupled with work-logic restructuring fosters clinical decision-making competence. Context endows knowledge with a value coordinate, anchoring it to the specific conditions under which a drug should be used or avoided. Feedback forges stable decision-making pathways through immediate, heuristic guidance that makes implicit reasoning explicit. Visibility transforms cognitive processes into objects that can be inspected, analyzed, and refined, fueling metacognitive growth. The progressive chain linking these mechanisms demonstrates that their coupling, rather than any single component alone, unlocks competence. The framework offers a generalizable diagnostic and design tool for AI-empowered curriculum reform across health professions education. ISRCTN Registry, ISRCTN15993331. Registered on 23 June 2026. Retrospectively registered.