SAHS Patients' Classification Based on Oximetry and Respiratory Effort Signal: An Alternative Method

摘要

The "Gold Standard" for Sleep Apnea/Hypopnea Syndrome (SAHS) diagnosis is the study of Polysomnography (PSG) in a sleep laboratory. It consists of connect to patient's body several sensors. The standards are the oronasal airflow (OAS) and pulse oximeter (SpO2) sensors while Respiratory Inductance Plethysmography sensor (RIP) is alternative. The airflow signal can be estimated from RIP signal (RIPFlow). Hypopneas are detected in %SpO2 desaturation events through baseline. However, there isn't a consensus about this value's definition. The signals sent by such sensors are analyzed by an expert to get the Apnea/Hypopnea Index (AHI) and to classify the patients into four groups: Normal, Mild, Moderate and Severe. In this study, an alternative method for scoring apnea/hypopnea events based on thorax and abdomen RIP sensor and for analyzing the %SpO2 values variations using Median Absolute Deviation (MAD) is proposed. For time domain comparison, the Pearson's correlation coefficient was computed on the RIPFlow with the respiratory flow signal. Also, the automatic algorithms from standard and proposed method were implemented to obtain AHI. In order to test the proposed method's performance, PSG recordings acquired in 23 adult patients are used. The Sensitivity (Sen), Specificity (Sp) and Accuracy (Acc) values were calculated considering patients classification for the standard method and well as the one proposed. Results indicate a high correlation (p-value < 0.05) in flow estimation and an improvement in patient classification using the model based on the RIPflow and the MAD-SpO2.