Cerebral stroke dysarthria risk prediction method based on ResNet and LSTM network

Abstract

The invention discloses a cerebral stroke dysarthria risk prediction method based on ResNet and a LSTM network. The method comprises the following steps: firstly, collecting voice information througha recording device, then preprocessing the voice information, and extracting voice feature parameters MFCC (Mel-scale Frequency Cepstral Coefficients); then constructing a ResNet and LSTM neural network model to train the MFCC feature parameters, and extracting depth feature information of a speech signal; and finally inputting the MFCC feature parameters to be tested into a trained model, and predicting the risk of the cerebral stroke dysarthria. The method has the advantages of being convenient, quick, cost-saving, high in prediction accuracy and the like.

Description

technical field [0001] The present invention relates to the technical field of deep learning, in particular to a method for predicting the risk of dysarthria in stroke based on ResNet and LSTM networks. Background technique [0002] "Cerebral stroke", also known as "stroke", is an acute cerebrovascular disease, a group of diseases that cause brain tissue damage due to sudden rupture of blood vessels in the brain or blockage of blood vessels that prevent blood from flowing into the brain. Patients will be unable to speak normally, aphasia symptoms of ambiguous expressions, and in severe cases, they will salivate involuntarily. According to the survey, stroke has become one of the biggest threats to the health of middle-aged and elderly people. Stroke will leave different degrees of limb dysfunction, sensory impairment, speech impairment, cognitive impairment, etc., and it is also the primary cause of disability in Chinese adults. According to investigations, among them, spee...

AI Technical Summary

Problems solved by technology

[0007] However, in the process of stroke risk prediction, these existing technologies need to collect relevant information for a long time, must have a large amount of case data, and the system prediction cycle is too long to meet the high efficiency standard

Using the traditional convolutional neural network as a stroke risk prediction model cannot guarantee that the accuracy rate will meet the standard

The above problems may delay the best treatment time, and there is uncertainty in risk prediction

Images