Patient similarity metric transfer system among disease domains on the basis of transfer learning

Abstract

The invention discloses a patient similarity metric transfer system among disease domains on the basis of transfer learning, and belongs to the technical field of the artificial intelligence software of computers. By use of the system, four submodules including a data preprocessing submodule, a similarity metric evaluation submodule, a similarity metric learning submodule and a similarity metric transfer submodule are constructed to finish similarity metric in a disease domain and patient similarity transfer among disease domains. The method aims to overcome the problem that traditional metric learning can not effectively work under a situation that supervisory information is difficult in acquisition and a patient sample amount in a specific disease field is small. The system can provide service support for a current accurate medical scene.

Description

technical field [0001] The invention relates to a patient similarity measurement system between disease domains based on migration learning, which belongs to the technical field of computer artificial intelligence software. Background technique [0002] With the advancement of the informatization of medical and health services, large medical institutions such as hospitals and physical examination centers have produced a large number of medical electronic health records. The data content mainly comes from the electronic medical records of hospitals, residents' health records collected by regional health information platforms, etc., and contains a large amount of unstructured / semi-structured data. It is an important task in the research of clinical decision support and patient group identification to give a clinically meaningful similarity measure between patients through outpatient, hospitalization, medication and health related data. Case query based on patient similarity c...

AI Technical Summary

Problems solved by technology

[0011] The purpose of the present invention is to overcome the problem that the existing similarity measurement system for three types of patients cannot work effectively under the condition that it is difficult to obtain supervisory information and the number of patient samples in a specific disease field is scarce, and to provide a method that can combine known source disease fields The measurement in the target disease field is migrated to the patient similarity measurement system in the target disease field, thus ensuring the smooth development of the research on the patient situation in the target disease field

Images