Jingjing Huang1, Liujie Ren1,2, Lifen Chen3, Zirui Jia3, Tianyu Zhang1,2, Haitao Wu4. 1. ENT Institute, Eye and ENT Hospital of Fudan University, Fenyang Road No. 83, Xuhui District, Shanghai, China. 2. FPRS Department, Eye and ENT Hospital of Fudan University, Shanghai, China. 3. Department of Aeronautics and Astronautics, Fudan University, Shanghai, China. 4. ENT Institute, Eye and ENT Hospital of Fudan University, Fenyang Road No. 83, Xuhui District, Shanghai, China. eentwuhaitao@163.com.
Abstract
PURPOSE: To explore the feasibility of automatic detection based on air flow and blood oxygen in patients with sleep disordered breathing. METHODS: This study proposes a new automated detection method for sleep disordered breathing based on overnight airflow and blood oxygen saturation (SaO2). In this regard, local range (LR) of the airflow was adopted to detect apnea events and the SaO2 sudden drops were used to help determine hypopnea events. Pearson correlation index was used to evaluate the relationship between the two automated methods (this study vs. Remlogic software) and the manual reports. Error and mean absolute error (MAE) were used to assess the two automated methods. RESULTS: For all patients, the apnea-hypopnea index (AHI), apnea index (AI) and hypopnea index (HI) for our automated scoring and manual reports were highly correlated (the Pearson correlation index were 0.996, 0.995 and 0.928, respectively, P < 0.001). However, HI for Remlogic automated scoring and clinical manual reports was poorly correlated (r = 0.316, P < 0.001). Compared with the manual reports, mean absolute error of AHI, AI and HI between the two automated methods (this study vs. Remlogic software) were statistically significant (P < 0.0001). Furthermore, among the three subgroups (group 1, AHI < 15/h, group 2, 15/h ≤ AHI < 30/h and group 3, AHI ≥ 30/h), the mean error and MAE of AHI between the two automated methods were also statistically significant (P < 0.01). CONCLUSIONS: Generally, good agreements were shown between our automated detection and clinical reports. This procedure is robust and effective, which would significantly shorten the analysis time.
PURPOSE: To explore the feasibility of automatic detection based on air flow and blood oxygen in patients with sleep disordered breathing. METHODS: This study proposes a new automated detection method for sleep disordered breathing based on overnight airflow and blood oxygen saturation (SaO2). In this regard, local range (LR) of the airflow was adopted to detect apnea events and the SaO2 sudden drops were used to help determine hypopnea events. Pearson correlation index was used to evaluate the relationship between the two automated methods (this study vs. Remlogic software) and the manual reports. Error and mean absolute error (MAE) were used to assess the two automated methods. RESULTS: For all patients, the apnea-hypopnea index (AHI), apnea index (AI) and hypopnea index (HI) for our automated scoring and manual reports were highly correlated (the Pearson correlation index were 0.996, 0.995 and 0.928, respectively, P < 0.001). However, HI for Remlogic automated scoring and clinical manual reports was poorly correlated (r = 0.316, P < 0.001). Compared with the manual reports, mean absolute error of AHI, AI and HI between the two automated methods (this study vs. Remlogic software) were statistically significant (P < 0.0001). Furthermore, among the three subgroups (group 1, AHI < 15/h, group 2, 15/h ≤ AHI < 30/h and group 3, AHI ≥ 30/h), the mean error and MAE of AHI between the two automated methods were also statistically significant (P < 0.01). CONCLUSIONS: Generally, good agreements were shown between our automated detection and clinical reports. This procedure is robust and effective, which would significantly shorten the analysis time.
Authors: Marcin Ciołek; Maciej Niedźwiecki; Stefan Sieklicki; Jacek Drozdowski; Janusz Siebert Journal: IEEE J Biomed Health Inform Date: 2014-05-23 Impact factor: 5.772