NIOSHTIC-2 Publications Search

Using voluntary cough characteristics to detect obstructive lung disease.

Day JB; Frazer DG; Goldsmith WT; Andrew M; Barkley J; Afshari AA; McKinney WG
Am J Respir Crit Care Med 2004 Apr; 169(7)(Abstracts):A201
Following standard pulmonary function testing at the West Virginia University Pulmonary Function Laboratory, volunteer patients were classified by physicians as either normal (men = 27; women = 25) or having obstructive lung disease (men = 27; women = 21). Three voluntary coughs were recorded for each subject using the system and procedure previously described (Goldsmith et al., Proc. 3rd Int. W. of Biosig. Interp.). A series of cough sound pressure wave and airflow analyses were conducted for each cough. The sound pressure wave analyses included the cough sound index (Goldsmith et al. Am J Respir Crit Care Med 157:A86 1998), octave filter analysis, determination of beta in a 1/f beta power spectral analysis and wavelet decomposition. Airflow signal analysis included measurements of peak flow, average flow, mean transient time, beta in a 1/f beta power spectral analysis, and flow pattern shape indices. A principal component analysis of all the data was performed. The 5 most significant components were selected as inputs to a quasi-Newton back-propagation neural network classification system. The neural network was trained with approximately half the subjects, then the remaining subjects were classified based on that training set. Results were used to construct ROC curves for men and women to compare the results of the cough analysis with the physicians' interpretation of their pulmonary function measurements. The sensitivity and specificity of the cough analysis method for men were equal at 0.93, and the area under the ROC curve, or test discrimination, was 0.97. For women, the sensitivity and specificity were equal at 0.72 and the test discrimination was 0.76. Results indicate that an accurate and rapid classification of patients with obstructive lung disease can be achieved using voluntary cough analysis.
Humans; Airway-obstruction; Airway-resistance; Pulmonary-system; Diagnostic-techniques; Diagnostic-tests; Respiratory-function-tests; Pulmonary-function; Pulmonary-function-tests
Publication Date
Document Type
Abstract; Conference/Symposia Proceedings
Email Address
Fiscal Year
Issue of Publication
NIOSH Division
Priority Area
Disease and Injury: Asthma and Chronic Obstructive Pulmonary Disease
Source Name
American Journal of Respiratory and Critical Care Medicine
Page last reviewed: September 2, 2020
Content source: National Institute for Occupational Safety and Health Education and Information Division