Prediction and Simulation of sleep stages based on EEG signals

Aiming at the problems of low efficiency and high misjudgment rate of sleep staging expert discrimination method, a sleep staging prediction model is constructed by using machine learning method. The model is mainly composed of three parts: data preprocessing, model prediction and experimental result analysis. Firstly, the EEG signals of five sleep points were standardized by Z-score; Then, KNN, lightgbm, multi classification SVM and random forest RF are used to study the prediction of sleep stages. Through the grid search method, the best classifier parameter combination is obtained in the training process and verified in the test set; Finally, four sleep stage prediction models are analyzed and evaluated by using six classifier indexes and five sleep stage confusion matrix. The accuracy rates of KNN, lightgbm, multi classification SVM and random forest RF classification models in the test set are 69.9%, 71.5%, 63.9% and 72.2% respectively. The results show that in the case of few data sets, the sleep stage prediction model based on classical multi classifier proposed in this paper can also be used to obtain a practical sleep automatic stage model.