Purpose/Significance To explore a semi-automated approach based on pre-trained language models，and to enhance both the efficiency and quality of medical exercise answer explanation generation. Method/Process A hybrid intelligence framework based on MC-BERT is introduced. Firstly，the question structure recognition，key knowledge point extraction and draft explanation formulation are automatically completed. Then，through manual verification，the accuracy and standardization of the content are strictly controlled to form a traceable reference explanation corpus. Result/Conclusion The method significantly improves generation efficiency，reduces manual intervention costs，while ensuring the consistency and traceability of the explanation content with the medical teaching syllabus and textbook knowledge system，providing a feasible path for the intelligentization of medical education.