Drug recommendation method and system for chronic obstructive pulmonary disease

Abstract

The invention discloses a drug recommendation method and system for chronic obstructive pulmonary disease, and belongs to the field of medical data processing. The method comprises the steps: converting each symptom into a vector by using a knowledge graph which takes symptoms and syndromes as entities and takes the relationship between the symptoms and between the syndromes as the relationship between the entities, constructing a symptom matrix, after inputting a drug recommendation model, obtaining drug vectors containing drug recommendation scores, and taking part of drugs with the highest scores as recommended drugs. The model comprises a feature extraction network used for performing feature extraction on the symptom matrix to obtain a symptom feature matrix; a first vector reshaping layer, which is used for reshaping the symptom feature matrix into a symptom feature vector; a recurrent neural network, which is used for decoding a tag vector from the symptom feature vector; and a scoring network, which is used for mapping the tag vector into a drug vector. According to the method, the influence of syndrome type prediction errors can be avoided while the relevance among symptoms, syndrome types and drugs is utilized, and the recommendation accuracy is improved.

Description

technical field [0001] The invention belongs to the field of medical data processing, and more specifically relates to a drug recommendation method and system for chronic obstructive pulmonary disease. Background technique [0002] Chronic obstructive pulmonary disease is a chronic respiratory disease with high morbidity and mortality. Frequent recurrences will cause patients to be unable to work, study and live normally, and have a serious impact on the quality of life of patients. Traditional Chinese medicine has shown significant advantages in the treatment of chronic obstructive pulmonary disease. However, due to the large base of patients with chronic obstructive pulmonary disease in my country and the relative shortage of traditional Chinese medicine diagnosis and treatment resources, the missed diagnosis of the disease is relatively common. impose a heavy economic burden. Therefore, it is very important to provide an intelligent diagnosis and treatment method and syst...

AI Technical Summary

Problems solved by technology

[0004] However, in the drug recommendation method disclosed in the above patent application documents, on the one hand, because it converts the patient's symptom information into a symptom feature vector through the operations of identifying text description, uniform variable unit, numericalization and normalization, the symptom The eigenvectors only contain the characteristics of each symptom itself, without considering the correlation between symptoms, which makes the accuracy of subsequent syndrome determination unguaranteed; on the other hand, the above-mentioned patent application documents strictly divide the drug recommendation into successive Syndrome type judgment and drug recommendation link, and the result of syndrome type judgment dominates the final drug recommendation result. Once there is a large error in syndrome type judgment, it will seriously affect the accuracy of drug recommendation results

Overall, the accuracy of existing drug recommendation methods for COPD needs to be further improved

Images