Sleep staging method based on deep learning

Abstract

The invention discloses a sleep staging method based on deep learning. The method comprises the following steps: performing oversampling on a single-channel electroencephalogram signal to obtain a data set; designing a convolutional neural network for sleep staging; pre-training is carried out on the data set after oversampling, and hyper-parameters of the model are adjusted according to Bayesianoptimization; designing a penalty weight loss function to perform secondary optimization on the model; and testing the input single-channel electroencephalogram signal by using the trained model to obtain a predicted sleep period. According to the method, the neural network can learn the sleep staging information without additionally extracting features, and the problem of low average recognitionrate of each period caused by imbalance of a data set can be effectively avoided by utilizing the method. The method can be widely applied to scenes with unbalanced data sets, such as electrocardiogram arrhythmia detection and electroencephalogram epilepsy detection.

Description

technical field [0001] The invention relates to the fields of pattern recognition and signal processing, in particular to a sleep staging method based on deep learning. Background technique [0002] Automatic sleep stage classification algorithms mainly include manual feature extraction and automatic feature extraction. Manual feature extraction methods extract features such as time, frequency, and time-frequency domain features from raw signals for training. Such methods may lose most of the original information since they only extract features. The method of automatic feature extraction can directly use the original data for training. Using the characteristics of end-to-end training of some neural networks, using them as feature extractors and classifiers at the same time can solve the limitations of manual feature extraction. [0003] The method of automatic feature extraction is currently using more algorithms, but the data set itself has a serious imbalance problem. M...

AI Technical Summary

Problems solved by technology

[0004] Purpose of the invention: In view of the above problems, the present invention proposes a sleep staging method based on deep learning, searches for suitable hyperparameters through Bayesian optimization, and uses a penalty weight loss function to optimize the model twice to solve the low average recognition rate of each sleep category The problem

Images