Receiver operating characteristic analysis of aerodynamic parameters obtained by airflow interruption: a preliminary...

Aerodynamic parameters provide objective and quantitative measures of laryngeal functional status. Jiang et al previously introduced an airflow interruption technique that can determine mean phonatory airflow, subglottic pressure, and phonation threshold pressure simultaneously and noninvasively. In this study, we performed receiver operating characteristic (ROC) analysis to evaluate this airflow interruption apparatus for voice profile analysis in subjects with laryngeal polyps or nodules or with Parkinson's disease. Patients with polyps (n = 14), nodules (n = 9), and Parkinson's disease (n = 12) were evaluated with the airflow interruption apparatus. Normal subjects served as controls. Mean airflow, subglottic pressure, and phonation threshold pressure were determined. Discriminant analysis was used to create linear equations combining all three parameters in order to obtain a new combined parameter. Solutions to the linear equations yielded values for the combined parameter that took into account mean airflow, subglottic pressure, and phonation threshold pressure. Combined parameter values were used as data to generate ROC curves. Laryngeal polyps were distinguished from the normal larynx with a sensitivity of 0.929 and a specificity of 0.933 at the point of maximal efficiency. The area under the ROC curve was 0.977 (Az). Patients with Parkinson's disease were distinguished from normal subjects with a sensitivity of 0.667 and a specificity of 0.909 at the point of maximal efficiency. The area under the ROC curve was 0.7958 (Az). Nodules were able to be distinguished from normal with a sensitivity of 0.889 and a specificity of 1.00. The area under the ROC curve was 0.9565 (Az). Our conclusions are twofold. First, aerodynamic parameters may be combined for simultaneous consideration by the construction of linear equations by means of discriminant analysis. Second, the airflow interruption apparatus, when used for voice function evaluation, has high sensitivity and specificity.