Parkinson's Disease Classification from /a/ Vowel : A Two Databases Comparison Study

Abstract

Parkinson's disease (PD) is a progressive neurological syndrome characterized by the degeneration of dopaminergic neurons responsible for movements. This decrease in dopamine causes the motor symptoms, such as tremors and muscle rigidity. Other non-motor symptoms may be observed such as speech impairments. Early detection of these vocal alterations in PD plays an essential role in its diagnosis. In this study, we propose a classification of Parkinsonian voice based on a comparative analysis of the vowel /a/, from two different databases. Through data visualization, we observed different statistical and acoustic characteristics for the same vowel. The results highlight an important conclusion: classifiers efficiency does not depend on the database used. Consistent classification performance was achieved while maintaining similar levels of accuracy, regardless of the specific characteristics of each dataset.