Detection and airflow obstruction severity grading of chronic obstructive pulmonary disease based on capnography

Chronic obstructive pulmonary disease (COPD) is a major chronic disease with high global mortality, and capnography offers a noninvasive method for testing lungs function under natural breathing. Reported here is the best paradigm for COPD diagnosis and airflow obstruction severity assessment based on capnography as achieved by comparing the performance of five machine learning methods. In this study, 1007 subjects underwent capnography and pulmonary function tests, then segment-fitted capnography was used to extract quantitative features. Using a hybrid scoring strategy to identify the head features, XGBoost performed the best, with accuracy, precision, recall, and F1 score of 90.08%, 90.33%, 90.08%, and 89.72%, respectively. This work shows for the first time the feasibility of using capnography to evaluate ventilation dysfunction in COPD patients. The proposed XGBoost-based method offers support to clinicians for assessing patients’ ventilatory function by capnography, which has considerable potential as an alternative to traditional pulmonary function tests for respiratory diseases.