Large-scale medical data knowledge mining and treatment scheme recommendation system

Abstract

The invention discloses a large-scale medical data knowledge mining and treatment scheme recommendation system. The system comprises a data set preprocessing module which is used for obtaining real electronic medical record data and carrying out preprocessing on the electronic medical record data composed of a plurality of types of heterogeneous data sources; a disease severity prediction module which is used for obtaining disease severity scores of each patient in the treatment process; a treatment effectiveness measurement module which is used for obtaining effective treatment measurement information; a patient similarity measurement module which is used for constructing a similarity measurement relationship of patients; and a drug therapy scheme recommendation module which is used for obtaining drug therapy scheme recommendation of the next stage. According to the invention, the severity degree of the illness state of the patient is judged and predicted through the multitask bidirectional heterogeneous LSTM, the effectiveness measure of treatment is defined, the fine grit similarity of the patient is calculated, and the treatment scheme of the next stage is recommended accordingto the historical treatment records of the patient and the effective treatment schemes of other patients with high pathology similarity.

Description

technical field [0001] The invention applies deep learning and knowledge introduction to realize the discovery and recommendation system of effective drug treatment plan, and belongs to the field of medical data mining. Background technique [0002] Electronic health record (EHR) data comes from millions of patients, and these data are currently collected and stored regularly in different medical institutions. These EHR data consist of heterogeneous data elements, typically including demographics, diagnoses, physical exams, sensor measurements, laboratory test results, prescribed or administered medications, and clinical records, among others. With the rapid development of information technology and the rapid popularization of electronic medical records (EMR), the amount of digital information stored in my country's electronic health records has surged in the past decade. It is generally believed that these massive data contain a lot of hidden knowledge, and various types o...

AI Technical Summary

Problems solved by technology

[4] mainly proposed three different LSTM deformations to solve the heterogeneity problem of data, but the overall framework of recommended treatment was not proposed in the article

Leilei Sun, Chuanren Liu et al. [5] proposed a data-driven automatic treatment plan development and recommendation method, mainly using the important information in the doctor's order, and the clustering method used finally obtained a few types of drug treatment combinations , unable to meet more refined treatment recommendations

At the same time, none of the above schemes takes into account the reactivity between drugs and the patient's history of drug allergy

Images