Applying composite physiological characteristics to assess the severity of obstructive sleep apnea

Obstructive Sleep Apnea (OSA) is a common sleep disorder which causes poor sleep quality, daytime drowsiness, memory loss, and lack of attention, and can contribute to accidents. Long-term OSA is associated with cardiovascular disease, high blood pressure, arrhythmia and diabetes. Early discovery and treatment of OSA can significantly reduce the difficulty and cost of treatment. OSA detection uses a scale method which is prone to subjective errors. Polysomnography (PSG) requires long detection times, entails equipment which can be difficult to obtain and high cost, and does not provide immediate diagnosis or the establishment of preventative treatment. Human biosignals have been found to provide a faster and more objective means of sleep apnea diagnosis, allowing for the early treatment of OSA. By identifying key physiological characteristics and observing their change over time, one can assess the severity of the patient's condition, and appropriate treatment can reduce the likelihood of disease onset or slow its progression. This study applies machine learning techniques to electroencephalography (EEG), electrocardiography (ECG), oxygen saturation (SpO(2)) and other complex physiological characteristics to establish a model for OSA severity and evaluate the effectiveness of various machine learning methods to detect sleep apnea. In addition, the fuzzy C-mean (FCM) algorithm is used to construct a predictive model for OSA to assist physicians in the early clinical detection of symptoms. The study found that patients with severe sleep apnea have a self-rescue condition which causes wakening due to respiratory arrest. Only relying on SpO(2)and the Apnea Hypopnea Index (AHI) will cause severe OSA cases to be misidentified as mild, and this study found that the A5 band brain wave can help detect sleep apnea, reducing the possibility of miscategorization and assisting physicians in accurate clinical diagnosis.