Transferable patient classifying system based on RNN neural network

Abstract

The invention belongs to the field of medical data mining, and discloses a transferable patient classifying system based on the RNN neural network. The system comprises an input module, a data processing module and a classifying module. The input module is used for acquiring original medical record data. The data processing module comprises a data preprocessing unit and a patient time sequence generating unit and is used for converting the original medical record data which are read by the input module to a patient time sequence. The classifying module comprises a model establishing unit, a model predicting unit, a migrating unit and is used for combining the RNN neural network with an attention mechanism for constructing a patient classifying model, then inputs the patient time sequence obtained by the data processing module into the model for training, and finally migrates the trained model to different data sets for performing patient classification prediction. The transferable patient classifying system has advantages of constructing the patient classifying model through mining the electronic medical record data, and assisting a doctor in diagnosis and treatment through performing patient death risk classification in different medical data sets.

Description

Technical field [0001] The invention relates to the field of medical data mining, in particular to a transferable patient classification system based on RNN neural network. Background technique [0002] As an index to measure the level of medical services, hospital mortality is of great significance in ICU care. The classification and prediction of ICU patients and the classification of ICU patients into whether they are high-risk patients can help formulate reasonable nursing plans for patients. If it is a high-risk patient, medical protection needs to be strengthened to reduce the risk of death. The higher the accuracy of patient classification, the more helpful it is for doctors to formulate scientific and reasonable treatment plans, thereby improving the quality of diagnosis and treatment, reducing ICU inpatient mortality, and improving the level of hospital medical services. Therefore, it is a valuable task to mine electronic medical record data and select appropriate char...

AI Technical Summary

Problems solved by technology

Perform data same-distribution processing on different medical datasets, so as to solve the problem that the same classification system cannot classify different medical datasets

Images