本研究透過紅外光脈動血氧計(Pulse oximetry)對受測者進行非侵入性的連續血氧含量偵測，針對心搏的光體積變化進行包含回歸分析與頻域分析等各項監測數據前處理，取得眾多特徵樣本後將此訓練集合(Train Set)輔以KNN分類演算法推估受測者臨床Respiratory/Disturbance Index (RDI)值，同時以物聯網架構將數據回傳至雲端，並透過無線通訊模組傳輸控制訊號與本研究研製的睡姿調整平台-安心枕連接。實驗亦設計多種的睡姿矯正觸發機制，綜合以上項目，經改良的安心枕期能有效的改善為睡眠呼吸中止所苦的人發生阻塞型呼吸中止之頻率以改善睡眠品質。;In this paper, we use non-invasive continuous detection to monitor blood oxygen and photoplethysmogram by pulse oximetry .For the various monitoring data preprocessing including regression analysis and frequency domain analysis are performed, and the training set is obtained after obtaining a plurality of feature samples. The KNN classification algorithm is used to estimate the clinical Respiratory/Disturbance Index (RDI) value of the subject, and the data is transmitted back to the Internet. The control signal is transmitted through the wireless communication module to the Sleep Apnoea Auxiliary Equipment ,which is developed by the research and called “POM Pillow”. This research also designed a variety of sleep posture for trigger condition. In conclusion, the “POM Pillow” can improve the effectively frequency of obstructive respiratory arrest in patients suffering from sleep apnea to improve sleep quality.
