Prediction of spirometric evaluations based on cough flow.
Goldsmith-WT; Reynolds-JS; Day-JB; Kashon-MK; Frazer-D
Am J Respir Crit Care Med 2013 May; 187(Meeting Abstracts):A1919
RATIONALE: Pulmonary function tests (PFTs) based on spirometric data are used extensively to assist in the diagnosis of lung disease. Many PFTs require the subject to perform a maximal forced expiratory maneuver which can be difficult to impossible for certain populations (infants, elderly, extremely sick). The purpose of this study was to determine if the airflow produced during a cough could be used to predict spirometric abnormalities. METHODS: Spirometery was conducted on subjects (n=107) in the pulmonary function laboratory of Ruby Memorial Hospital. The FEV1, FVC and FEV1/FVC values for each subject were calculated and were considered normal if they were greater than the lower limits determined by Hankinson et al. (Am J Respir Crit Care Med. 159(1):179-187, 1999). Airflow patterns generated during voluntary coughs were then recorded using a unique system that could simultaneously measure cough sounds and airflow properties (Goldsmith et al., Ann of Biomed Eng. 38(2): 469-477, 2010). Twenty-one features were calculated from the cough flow signal (peak flow, volume, etc.) and four features were based on subject demographics (weight, height, age and gender). Feature selection and a support vector machine were implemented within a genetic algorithm to classify subjects as having either normal or abnormal FEV1 , FVC and FEV1/FVC values based on their coughs. A double nested cross-validation technique was used to evaluate the prediction accuracy of this type of classifier. RESULTS: The accuracy of the classifier was 78%, 65%, and 80% for FEV1, FVC and FEV1/FVC values, respectively. CONCLUSION: Cough flow signals contain information that can be used to predict whether a subject will produce abnormal FEV1, FVC and FEV1/FVC measurements during spirometery.
Pulmonary-function-tests; Spirometry; Respiratory-system-disorders; Pulmonary-system-disorders; Lung-disease; Airway-obstruction; Vital-capacity; Diagnostic-tests; Humans; Air-flow; Equipment-reliability; Performance-capability; Measurement-equipment; Diagnostic-techniques; Analytical-instruments
W. T. Goldsmith, National Institute for Occupational Safety and Health, Morgantown, WV
American Journal of Respiratory and Critical Care Medicine