Nonlinear glottal flow features in Parkinson's disease detection

As a methodology for automatic detection of Parkinson's disease (PD), it is proposed the estimation of the different glottal flow features considering nonlinear behavior of the vocal folds. This paper evaluates the discrimination capability of set with eight different Nonlinear Dynamic (NLD) features. The experiment presented considering the five Spanish vowels uttered by 50 People with PD (PPD) and 50 Healthy Controls (HC). According to the results, it is possible to achieve accuracy rates of up to 75.3% when only the vowel |e| is considered.