Multivariate Causality Analysis Method Based on Adaptive Selection of Lag Order

Abstract

The invention relates to a multivariate causality analysis method based on adaptive selection of lag orders, aiming at accurately predicting the causal influence between various brain regions based on motor imagery EEG signals. At present, the traditional Granger causality method based on autoregressive model lacks consideration of the lag-dependent structure existing in multivariate time series and the influence of model coefficients on causality. This project first obtains multi-channel motor imagery EEG signals, and then uses the improved backward time selection algorithm to estimate the optimal lag order of each variable in the regression model, establishes a dynamic regression model of multi-channel EEG signals, and then uses the model's Residuals and coefficients are used to define conditional causality measures between multiple variables, which can effectively improve the estimation performance of true causality. This method has broad application prospects in the fields of causal-effect brain function network and cortical-muscle coupling analysis.

Description

technical field [0001] The invention belongs to the field of brain function network analysis, and relates to a method for self-adaptive selection of lagging order of each variable in a dynamic regression model and a multivariate causal relationship analysis method. Background technique [0002] The human brain is one of the most complex dynamical systems in the world. Its cortex is composed of 15-33 billion nerve cells. The interconnections between these neurons form a brain network with a fairly complex structure and function. In terms of physiology and physiology, the main function of the brain is to control and dominate various organs of the body. Therefore, this complex and huge brain network enables it to have advanced information processing and cognitive expression functions, such as: language, emotion, memory, cognition, etc. , and store, process, process and integrate information from the inside of the human body and the surrounding environment, explore the structure...

AI Technical Summary

Problems solved by technology

However, in the traditional GC analysis method based on the linear regression model, all variables in the linear regression model have the same lag order, which does not take into account the time-lag dependence structure that usually exists in multivariate time series, which limits the estimation performance of the model

Images