Nonlinear Dynamic Analysis of Disordered Voice: The Relationship Between the Correlation Dimension (D2) and Pre-/Post-Treatment Change in Perceived Dysphonia Severity

Abstract

The purpose of this study was to evaluate the clinical utility of nonlinear dynamic analysis methods, including phase space portraits and measures of the correlation dimension (D(2)) to predict pre- versus post-treatment change in perceived dysphonia severity in a group of 88 patients with muscle tension dysphonia (MTD). Pre- and posttreatment vowel samples from 88 women with primary MTD (mean age=46.2 years; standard deviation=13.1) were selected for analysis (176 voice samples in total). Phase space reconstructions and correlation dimensions were computed to describe the nonlinear dynamic characteristics of all voice samples. Ten blinded listeners were asked to rate the vowel samples for severity of dysphonia using a 100-point visual analog scale (VAS). In the computation of D(2) results, 22 severely dysphonic pretreatment voice samples were not analyzed, as a finite value for the correlation dimension could not be computed. For the remaining pre-/post-treatment voice samples, a significant difference in the correlation dimension (D(2)) between the pre- versus post-treatment voice samples was observed; however, D(2) was poorly correlated with changes in perceived dysphonia severity ratings after treatment (r=0.244, P=0.056). Thus, the utility of the correlation dimension (D(2)) as a treatment-outcome measure and as a measure of dysphonia that may strongly relate to perceived dysphonia severity does not appear to be supported, particularly for pretreatment voices that may have increased levels of dysphonia severity. Instead, the strength of nonlinear dynamic methods may potentially reside in providing some insight into the theoretical rules or initial conditions that may result in different modes of normal or disordered phonation.