To explore and implement the automated interpretation of pediatric medical exam answers, so as to enhance the efficiency and quality of answer explanation compilation. Method/Process The paper proposes a method that combines latent semantic indexing (LSI), MC-BERT, and the CoSENT model. Initially, multiple candidate answer explanations are extracted from reference documents using the LSI method and the MC-BERT model. Subsequently, the CoSENT model is employed to calculate the similarity between the candidate explanations and the question stems as well as the answer options. The candidate explanation with the highest similarity is then selected as the final answer explanation. Result/Conclusion The experimental results show that the method presented in this paper achieves a precision rate of 72.6%. Compared to single methods or models, it significantly improves the recall and precision of answer parsing, effectively enhances the efficiency of compiling question answer explanations, reduces the burden on educators, and provides significant data support for educational research.