Causal medical diagnosis method based on deep learning

Abstract

The invention discloses a causal medical diagnosis method based on deep learning. The method comprises a diagnosis and treatment model based on a deep neural network and a causal learning module with knowledge regularization processing. The diagnosis and treatment model based on the deep neural network utilizes the efficient nonlinear learning ability of the deep neural network to obtain the deep features of the complex physiological signals, and the causality learning module combines knowledge regularization to carry out causality discovery. The method has the beneficial effects that 1) the deep neural network is utilized to obtain deep features, and then variable pairs are formed by the features and diseases, so that the causal relationship of disease characterization is further mined; 2) a causality learning module with knowledge regularization processing is realized by using a neural network and can be embedded into any current automatic diagnosis and treatment model, and the whole method can be learned and optimized by using a gradient descent method; and 3) a knowledge-based regularization technology is adopted, and a neural network classification result is improved in combination with axiom and domain knowledge, so that causal discovery is more stable.

Description

technical field [0001] The present invention relates to the technical field of machine learning, and more specifically, relates to a method for discovering causality of medical events that combines deep learning and knowledge regularization. Background technique [0002] Health data such as from electronic medical records, intensive care unit data streams, and patient-generated health data are becoming more widespread and have the potential to be used to discover the causes of disease. However, discovering diseases from physiological data is a big challenge, because human physiology is complex and non-linear, and multiple diseases often share similar symptoms and physiological manifestations. For example, the root cause of acute respiratory distress syndrome (ARDS) is respiratory system acute failure, but may present with circulatory or neurologic symptoms. Without modeling and analyzing this complexity, it is impossible to truly explore the nature and laws between diseases...

AI Technical Summary

Problems solved by technology

Without modeling and analyzing this complexity, it is impossible to truly explore the nature and regularity between diseases and symptoms

[0003] In recent years, with the ability of deep learning-based artificial intelligence technology to model complex relationships from input to output, it has achieved great success in many application fields, but sometimes it is easy to overfit the irrelevant parts of the data. Other features, excessive attention to secondary connections from cause to effect, or even occasional irrelevant connections, thus reducing the fitting of the essential relationship of the problem

Images