Identification of Asthmatic Patient During Exercise Using Feature Extraction of Carbon Dioxide Waveform

Asthmatic athletes encounter greater challenges while practicing sport in dealing with their disorder. Hence, this study explores the significance of features extracted from the different phases of carbon dioxide (CO2) waveform morphology from an asthmatic patient during exercise. Herein, CO2 data was collected using human respiration CO2 measurement device from 9 asthmatic subjects in the stable mild state before and after exercise, aged between 20–25 years via convenience sampling method, chosen from the UTM Health Center. The subjects were asked to run on the medical treadmill (TMX428-15% elevation, 7.5 m/h) for 2 minutes. Thereafter, we automatically segmented each breath cycle into sub-cycles using threshold and computed Area as a feature from each part using numerical methods compared with slope or derivative of the capnogram. We found that the feature (Area) for the segmented part of the mix of upward expiratory phase and alveolar phase possess higher area under curve (AUC) of 0.94 via receiver operating curve analysis. Further, the feature was sent to the classifier for the classification of athlete asthmatic condition before and after exercise. We found that the decision tree possesses higher accuracy (88.89%), sensitivity (100%) and specificity (77.78%) with an AUC of 0.85 than other classifiers. Thus, the incorporation of these features into a newly developed device could possibly allow asthmatic athletes to manage their condition during exercise and sports.