Automated Acoustic Analysis in Parkinson's Disease Using a Smartphone.

Abstract

Dysarthria is a common speech disorder in Parkinson's Disease (PD). The Dysarthria Analyzer software has emerged as a viable tool for automatic speech analysis in PD and quantification of dysarthria severity. However, most studies use the Dysarthria Analyzer with recordings obtained under tightly controlled conditions and high-quality microphones, and the utility of the Dysarthria Analyzer when used with recordings acquired under non-ideal conditions, such as in busy clinical settings, remains unexplored. This study investigates the Dysarthria Analyzer's performance in a setting more akin to a clinical environment using a smartphone. We obtained data from three groups, including healthy controls (HC), PD patients with their deep brain stimulation on (ON-DBS), and PD patients with their DBS off (OFF-DBS). We found a significant decrease in pitch variability and an increase in speech rate for the OFF-DBS group compared to the HC. Furthermore, most of the estimated values for the speech markers fall within the reported values in the literature. Our findings demonstrate that the Dysarthria Analyzer effectively extracts relevant speech markers even when used with recordings obtained under non-ideal conditions, emphasizing its potential for widespread clinical adoption.Clinical Relevance- Our findings demonstrate the potential of using smartphone recordings obtained in clinical environments for automatic objective speech analysis. These findings are relevant for developing a clinical tool that can be widely accessible and easily implemented during routine clinical visits of PD to improve the assessment of dysarthria in PD.