目的：職能治療臨床推理是臨床專業能力的關鍵，但對無臨床經驗的學生而言具挑戰性。而生成式人工智慧技術如 ChatGPT 提供了低成本且靈活的解決方案，能生成虛構情境，避免患者隱私和倫理問題，並具有潛力成為職能治療領域的創新教學工具，以提升學生臨床準備能力；因此本研究擬探討課程中使用 ChatGPT 輔以教學工具下的學生使用情形。 方法：研究對象為 33 名大學三年級職能治療學生，於課程中操作 ChatGPT 並輔以「臨床推理個案生成指引表」與「提示自我評量表」進行學習。課程前後以問卷蒐集學生使用經驗，並以 Wilcoxon 符號檢定分析其變化。結果：問卷分析顯示，學生在「生成不同個案與情境」的操作可用性顯著提升（p = 0.028*），整體面向亦達顯著差異（p = 0.03*），顯示學生逐漸熟悉操作並降低感知困難度。在「生成評估報告書」部分，雖整體未達顯著差異，但「相符性」題項呈現顯著下降（p = 0.042*），反映學生對生成內容的一致性要求更嚴格。在「生成療育建議書」部分則未呈現顯著變化。在可用性、合理性、相符性、協助性四項使用感知指標中，均未達統計顯著差異。結論：生成式工具 ChatGPT 輔助課程能有效支援學生生成臨床個案與情境，降低操作門檻並促進臨床推理學習，具備低成本、靈活性高且避免倫理問題的優勢，此項方法在職能治療教學中的應用具備潛力，未來仍需結合更多情境與跨校研究以驗證其成效。

Objective: Clinical reasoning is a key professional skill in occupational therapy, but it is challenging for students without clinical experience. Generative AI tools, such as ChatGPT, have the potential to become innovative teaching tools in occupational therapy, enhancing students’ clinical preparedness. This study aims to explore students’ use of ChatGPT in a teaching environment supplemented with specific instructional tools. Methods: Thirty-three third-year occupational therapy students participated in a course that integrated ChatGPT with structured instructional tools, including the Case Generation Guidelines and a Prompt-Based Self-Evaluation Form. Students completed a 12-item Likert-scale questionnaire before and

after the course. Data were analyzed using the Wilcoxon signed-rank test. Results: Survey analysis revealed a significant improvement in usability for generating cases and scenarios (p= 0.028*), with the overall domain also reaching significance (p = 0.03*), indicating increased familiarity and reduced perceived difficulty. For evaluation reports, no overall significant

difference was observed; however, the consistency item showed a significant decline (p = 0.042*), suggesting stricter standards applied by students. No significant change was found in

generating intervention plans. Across the four perception indicators—usability, reasonableness, consistency, and supportiveness—no statistically significant differences were detected.

Conclusion: ChatGPT shows promise in supporting case generation and clinical reasoning by lowering barriers and avoiding ethical concerns. Its cost-effectiveness and flexibility highlight potential for occupational therapy education, though broader validation is needed.