Research On The Generation Method Of TCM Clinical Diagnosis Results Based On Long Short-term Memory Coding And Decoding Framework

The unbalanced development of medical resources renders the research of clinical assistant diagnosis of Traditional Chinese Medicine(TCM)particularly important in areas where medical resources are relatively scarce.TCM clinical assistant diagnosis refers to the process of using TCM clinical diagnosis data,through calculation and analysis,to establish a clinical diagnosis model for predicting diagnosis,and provide medical staff with diagnostic references.It will assist the clinical diagnosis process of TCM and improve the efficiency and quality of TCM clinical consultation.This article mainly studies relevant problems of TCM clinical assistant methods.The main research focus are as follows:(1)Data representation of TCM clinical assistant diagnosis: Traditional data representation uses vector space representation,which is extremely susceptible to dimensional disasters,and the semantic representation is not accurate enough to meet the rich semantic representation requirements of clinical consultation data of traditional Chinese medicine.In this paper,the distributed characterization method is utilized to conduct relevant experimental research.Among them,the fast-Text characterization improves the prediction efficacy of the traditional model,which is suitable for this task.(2)Research on multi-label classification of clinical syndromes of TCM based on text generation: The traditional method of TCM clinical diagnosis treats the task as a multi-label classification task,and obtains the top-n result of the predicted diagnosis.In this paper,a multi-label classification method based on text generation is used to complete the clinical diagnosis task of TCM.Its advantage lies in the dynamic generation of labels.The number of labels is determined by the model's prediction.This approach is more accurate and flexible.Using the same data representation methods,the experimental results prove that the multi-label classification prediction model of TCM clinical syndrome based on text generation is superior to traditional models,which improves the predicting ability of the diagnostic model.(3)Research on the generation of clinical diagnosis results of TCM based on text generation: The multi-label classification method based on text generation is not sufficient for solving the problem of unknown syndrome description.Starting from the theory of TCM,this paper replaces the original label level output semantic unit with word level and character level ones,that is,replaces the syndrome description with "symptom element" as the minimum semantic unit.The experimental results prove that the syndrome elements is able to complete the task of generating clinical diagnosis results of TCM and solve the problem of unknown syndrome description.