Single-channel electroencephalogram signal sleep staging method and system based on integrated model

Abstract

The invention provides a single-channel electroencephalogram signal sleep staging method and system based on an integrated model. The single-channel electroencephalogram signal sleep staging method based on the integrated model comprises the steps that electroencephalogram signals of a single channel of a patient and the characteristics of a subject are collected; the obtained single-channel electroencephalogram signals are preprocessed, and the complete single-channel electroencephalogram signals are divided into a plurality of segment signals with a certain duration; a plurality of sectionsof rhythm waves in each segment signal are obtained; time domain characteristic parameters of each segment signal and frequency domain characteristic parameters of each rhythm wave on each segment signal are extracted; and the obtained characteristic parameters and user characteristics are input into the constructed integrated model to obtain a sleep staging result of the single-channel electroencephalogram signals, and the integrated model is established based on an improved random forest algorithm and an improved LightGBM algorithm and gives an adjustable weight penalty factor to each sleepperiod.

Description

technical field [0001] The present invention relates to the field of sleep staging, and in particular to a method and system for sleep staging of single-channel EEG signals based on an integrated model. Background technique [0002] Sleep is one of the most important circadian rhythms in human physiological activities, and its quality affects our daily behaviors, such as learning, memory, and attention. When sleep deprivation continues, it often leads to diseases such as high blood pressure, sleep apnea syndrome, obesity, cardiovascular disease, Alzheimer's disease and Parkinson's disease. Existing scientific studies have shown that the physiological signals recorded by electroencephalogram (EEG) are beneficial to analyze the changes of sleep cycles, and are of great significance for the diagnosis and treatment of sleep-related diseases. [0003] For the study of sleep staging, the collected EEG is generally divided into 30s segments, and each segment is divided into five p...

AI Technical Summary

Problems solved by technology

These studies only considered the EEG signal characteristics and ignored some characteristics of the subjects themselves, such as age and gender, which affected the recognition rate of sleep stages to a certain extent.

In addition, due to some characteristics of sleep itself, the distribution of data fragments in each period is very different, especially in the N1 period, which is relatively small, which leads to a large deviation in the machine learning algorithm, which in turn makes the recall rate of the N1 period very low

Images