A system that recommends diagnostic cases by deducing the degree of similarity using the artificial neural network technique for the patient's main symptom and diagnostic relationship

摘要

The present invention relates to a system for inferring the level of similarity through an artificial neural network (ANN) technology used for relationship between a chief complaint (CC) of a patient and diagnosis to recommend a diagnosis case, which recommends a diagnosis case (hereinafter, including a detailed disease and prescription) about the corresponding CC through the ANN technology after a clinician inputs text about the symptom during a medical interview of the patient. According to the present invention, the system comprises: a first step of allowing the clinician to input text about a symptom to an electronic medical record (EMR) during a medical interview; a second step of analyzing atypical text data to extract a term (representation) about CC; a third step of setting diagnosis result data about the patient as a training dataset for each CC and setting a part of the data as a verification dataset and a test dataset; and a fourth step of analyzing a diagnosis pattern through training, which uses an unsupervised deep feature extraction learning method for the relationship between the CC and the diagnosis, to extract a feature from the training dataset.