Evaluation of DeepSeek-Generated Biochemistry Clinical Cases for Teaching Purpose

To overcome limitations in traditional case-based teaching, this study developed an AI-driven workflow using DeepSeek to generate contemporary, interdisciplinary clinical biochemistry cases. Through human-AI collaboration, eight structured cases covering key topics such as carbohydrate and lipid metabolism were created, each including a clinical description, molecular mechanisms, and Q&A, followed by instructor review. Fifteen medical students and 15 instructors evaluated the cases using a 5-point Likert scale across multiple dimensions, while student performance was compared between a group using AI-generated cases and a control group. The AI generated each case in 10-15 min, significantly faster than manual development. Cases presented a logical progression from molecular mechanism to clinical management and incorporated recent advances such as CRISPR and PCSK9 inhibitors. Content integration received the highest ratings, though instructors scored pedagogical applicability lower. Error analysis indicated that AI excelled in maintaining logical consistency, whereas human reviewers enhanced precision in clinical details. Students who used the AI-generated cases achieved significantly higher examination scores than the control group (p < 0.05). In conclusion, DeepSeek-generated cases are efficient, interdisciplinary, and innovative. Human review remains essential for ensuring clinical rigor, particularly in nuanced scenarios. This collaborative approach enhances both the efficiency of case development and the educational quality of biochemistry teaching materials.