Method for Detecting Subliminal Parkinsonian Deficits based on Automatically-analyzed Speech Features Data

Abstract

Parkinson's disease (PD), one of the most prevalent and well-known neurodegenerative disorders, has raised great concern in society. For many years, sensory testing has been the only method used to assess speech deficits in neurodegenerative diseases like PD. Therefore, by mining an automatically-analyzed speech dataset, a model was developed and an attempt was made to separate PD patients and RBD patients from general population. The acoustic features involved in this study included timing, articulation, phonation, and respiration, which were collected and analyzed based on recordings of participants reading a passage. The model was able to identify the patient's speech deficits by recognizing variations between the samples. According to this study, patients with related neurodegenerative disorders could suffer from a certain pattern of speech abnormalities that enables them to behave differently from healthy controls. This approach is intended to provide researchers with thoughts for future research into speech deficits and possible treatments associated with neurodegenerative illnesses.