Composite disease clinical path construction method and system based on transfer learning

Abstract

The invention relates to a composite disease clinical path construction method and system based on transfer learning. The method comprises the following steps: acquiring multi-source heterogeneous medical data of a composite disease generated by a non-clinical path; structuring the data according to ICD classification to form a first data set; performing feature extraction and clustering on the first data set according to the sign information, the medical staging, the diagnosis information and the treatment evaluation to form a second data set; randomly extracting data sets of two disease types from the second data set, training a time domain convolutional neural network to learn one disease type data set, and adjusting the time domain convolutional neural network through MMD to be used for transfer learning of the other disease type data set; and finally, fusing according to an output result to obtain a composite disease clinical path. By simplifying the data set and integrating the data set with the time domain convolutional neural network and the generative adversarial neural network, the performance and interpretability of a composite disease clinical path structure are improved, and gradient explosion is avoided.

Description

technical field [0001] The present invention relates to the field of medical information and deep learning, in particular to a method and system for constructing clinical pathways of complex diseases based on transfer learning. Background technique [0002] Clinical Pathway (Clinical Patlhmays, CP) refers to a disease or surgical operation corresponding to a certain International Classification of Diseases (ICD), based on evidence-based medicine, with the purpose of expected therapeutic effect and cost control A procedural and standardized diagnosis and treatment plan formulated with strict work order and accurate time requirements. [0003] Since the development of clinical pathways in my country is in the initial stage, the formulation of clinical pathways mainly relies on traditional expert evaluation, which takes a long time, costs high, and has a large degree of variability in clinical practice. So far, there are only about 1,200 clinical pathways formulated by the Nat...

AI Technical Summary

Problems solved by technology

[0010] The present invention aims at the technical problems of lack of timing, too deep model layers, weak generalization, large amount of calculation, and poor interpretability existing in the existing clinical path construction technology for complex diseases. The method for constructing clinical pathways for compound diseases includes the following steps:

Images