With the rapid growth of teaching resources, the sharing efficiency of teaching resources in traditional Chinese medicine courses in universities is low. This problem has not been solved. A design and research on a machine learning based model for sharing teaching resources in traditional Chinese medicine has been proposed. Using unified standards to construct the ontology of traditional Chinese medicine teaching resources, based on this, a retrieval tool is used to retrieve the teaching resources of traditional Chinese medicine courses, and a teaching resource classifier is constructed using machine learning technology - BP neural network. Divide the teaching resources of traditional Chinese medicine and pharmacy courses into multiple categories. Based on the classification results of teaching resources and the sharing needs of teaching resources in traditional Chinese medicine courses, a teaching resource sharing model is constructed to achieve the sharing of teaching resources in traditional Chinese medicine. The experimental data shows that the minimum retrieval time of the teaching resources obtained after applying the construction model is 0.3 s, and the maximum sharing efficiency of teaching resources is 960 MB/s, fully confirming the good application performance of the construction model.