Early diagnosis method and system for cerebral infarction based on electroencephalogram and artificial intelligence

Abstract

The invention provides a cerebral infarction early diagnosis method and system based on an electroencephalogram and artificial intelligence, and relates to the technical field of cerebral infarction evaluation.The method comprises the steps that high-density electroencephalogram signals of a patient are collected, multi-scale signal components are obtained and converted into symbol sequences, and a complexity index set is obtained; an entropy sequence of the signal on different time scales is calculated to obtain a multi-scale irregularity index, and the complexity index set and the multi-scale irregularity index are integrated to obtain an electroencephalogram feature tensor; inputting the electroencephalogram feature tensor into a space-time diagram convolutional network, modeling a time-varying connection relationship between brain intervals, dynamically adjusting a topological structure, generating a time-varying brain function connection diagram, finally outputting a cerebral infarction risk score, and generating an interpretable thermodynamic diagram; based on the cerebral infarction risk score and the interpretable thermodynamic diagram, a cognitive calculation framework is constructed, and the cognitive calculation framework obtains a state transition probability matrix; generating a causal atlas; calculating an intervention effect score; and generating a diagnosis report.

Images