Research On Key Method For Traditional Chinese Medicine Prescription Recommendation Based On Deep Learning

Under the strategic background of "Healthy China",the development of traditional Chinese medicine(TCM)shoulders important historical missions and responsibilities in this era.TCM clinical diagnosis and treatment have always relied on the subjective medical experience judgment of TCM physicians.TCM case records,as the main presentation form of TCM physician’s diagnostic process and results,contain rich diagnostic and therapeutic experiences and abundant medical theories of TCM physicians.With the development of artificial intelligence technology,deep learning is widely applied in the field of TCM.This technology can excavate diagnostic and therapeutic experiences from TCM case records,and assist TCM physicians in clinical decisionmaking.This article applies deep learning technology to assist TCM diagnosis and treatment by analyzing and studying relevant technologies,summarizing TCM text characteristics,researching TCM assisted diagnosis methods,and implementing TCM prescription recommendation based on the collected TCM case record data.The work of this article mainly consists of the following two aspects.:(1)For the task of disease-aided diagnosis,this paper proposes a TCM-aided diagnosis technique for discovering hidden knowledge.Firstly,the TCM text is pre-processed and then the text representation is obtained.Secondly,the text representation is input into the ERNIE layer to generate word vector representation using the self-attentive bidirectional modeling capability of Transformer,so as to obtain the contextual information of the input text.The vectors obtained by ERNIE layer are input to LSTM layer,and the text features are further extracted by LSTM.Finally,the final feature vectors obtained from the LSTM layer are The proposed TCM disease-aided diagnosis method in this paper performs better than the comparison methods.It uses the fullyconnected layer for classification and obtains final prediction results of the model.Experimental results support these findings.(2)For the task of clinical prescription of traditional Chinese medicine,this thesis proposes a recommended prescription model based on Seq2 Seq model,It is difficult to fit the complex relationship between traditional Chinese medicine and syndrome types and symptoms.To solve this problem,this thesis uses the Attention mechanism to improve the sequence-to-sequence generation model and learn the correlation of symptoms,syndrome types and prescriptions in sentences.In addition,in view of the particularity of traditional Chinese medicine text,this thesis adds an explicit n-gram mask language model to further increase the semantic understanding ability.The experimental results show that the TCM prescription recommendation model proposed fenfprescriptions during clinical diagnosis and treatment.In summary,this article mainly builds a model for prescription generation in clinical patient visits by using the ERNIE model and sequence-to-sequence method in deep learning.It aims to achieve auxiliary diagnosis and prescription recommendation in traditional Chinese medicine,and verify the feasibility of the method through experiments.This provides a theoretical basis for assisting traditional Chinese medicine physicians in clinical diagnosis and improves the quality and efficiency of traditional Chinese medicine clinical visits.