Classification of voice signals to aid in the diagnosis of Parkinson’s disease

Authors

  • Mateus Melo IFPB
  • Thiago Gouveia Instituto Federal da Paraíba (IFPB)

DOI:

https://doi.org/10.5335/rbca.v15i2.13556

Keywords:

Machine learning, K-fold cross validation, parkinson's disease, random Forest

Abstract

The Parkinson’s disease is a neurodegenerative illness which impairs motor and speech skills, in addition to provoke
behavior, mood and thinking changes. It hits, more usually, the elderly population and its diagnosis is done by a clinical
exam, by the observation of a patient’s symptoms. Since the most notorious symptoms appear in advanced stages of
the disease, which makes the treatment more difficult, and that the world is passing through a process of age pyramid
inversion, the tendency is that the Parkinson will become a global public health issue. Within this context, proposals for
the use of voice signals as a form of early diagnosis of Parkinson’s have been achieving results. This work proposes the
construction of a tool to aid in the diagnosis of Parkinson’s disease using voice signals associated with Machine Learning
techniques. Using a set of data with different attributes extracted from the speech of carriers and non-carriers of the
pathology, an accuracy of 93.8 % was obtained using a Random Forest algorithm and performing a cross-validation with
the k-fold technique.

Downloads

Download data is not yet available.

Published

2023-07-27

Issue

Section

Original Paper

How to Cite

[1]
2023. Classification of voice signals to aid in the diagnosis of Parkinson’s disease. Brazilian Journal of Applied Computing. 15, 2 (Jul. 2023), 88–104. DOI:https://doi.org/10.5335/rbca.v15i2.13556.