Abstract
Objective: Simple screening questionnaires have been designed for detecting Obstructive Sleep Apnea (OSA) in the general population but may not be suitable for COPD patients. In this study we developed a simple questionnaire by the adaptive network-based fuzzy inference system (ANFIS) to detect the coexistence of severe OSA in patients with COPD.
Methods: Seventy-eight of 287 COPD patients was enrolled as the training group. They received polysomnography and were asked if they have snoring and witnessed apnea. Twelve were excluded because of central sleep apnea. We proposed an ANFIS to predict the severity of OSA based on the three factors from the patients of the training group: BMI, snoring and witnessed apnea. After the training process of ANFIS completed, we consecutively enrolled 67 COPD patients as the validation group. All of them received polysomnography and answered the same questions. Then we applied the trained algorithm of ANFIS to the three factors for predicting the severity of OSA in the validation group.
Results: There were significant correlations between the predicted and measured AHI in the training and validation group (r=0.7312, 95% CI 0.5943-0.8269, p<0.0001 and r=0.8029, 95% CI 0.6971-0.8745, p<0.0001, respectively). To detect the severe OSA in the validation group, the ANFIS-based algorithm provided a high accuracy with a sensitivity of 90.9%, specificity of 95.6%, positive predictive value of 90.9% and negative predictive value of 95.6%.
Conclusion: ANFIS is a useful tool to develop a simple questionnaire for detecting the severe OSA in COPD patients.
- © 2014 ERS