Parkinson disease classification using one against all based data sampling with the acoustic features from the speech signals.

Parkinson's disease (PD) is a long-term degenerative disease that primarily affects the motor system of the central nervous system. This disease is difficult to diagnose and is one of the common diseases in the public. In this paper, we have proposed a novel data sampling method for the classificati...

Descripción completa

Detalles Bibliográficos
Publicado en:Medical Hypotheses Vol. 140
Autores principales: Polat, Kemal, Nour, Majid
Formato: Journal Article
Publicado: Elsevier B.V. Jul2020
Acceso en línea:Ver este registro en EBSCOhost
fields @attributes:
  recordID: 1
pdfLink:
plink: https://search.ebscohost.com/login.aspx?direct=true&db=ccm&AN=143703133&site=ehost-live
header:
  @attributes:
    shortDbName: ccm
    uiTerm: 143703133
    longDbName: CINAHL Complete
    uiTag: AN
  controlInfo:
    bkinfo:
    dissinfo:
    jinfo:
      jid:
        03069877
        NZO
      jtl: Medical Hypotheses
      issn: 03069877
      maglogo: N
    pubinfo:
      dt: Jul2020
      vid: 140
      pid: 467
      pub: Elsevier B.V.
      place: New York, New York
    artinfo:
      ui:
        143703133
        143703133
        NLM32197120
        10.1016/j.mehy.2020.109678
        NLM32197120
        143703133
      ppct: 1
      formats:
      tig:
        atl: Parkinson disease classification using one against all based data sampling with the acoustic features from the speech signals.
      aug:
        au:
          Polat, Kemal
          Nour, Majid
        affil: Department of Electrical and Electronics Engineering, Faculty of Engineering, Bolu Abant Izzet Baysal University, 14280 Bolu, Turkey
      sug:
      ab: Parkinson's disease (PD) is a long-term degenerative disease that primarily affects the motor system of the central nervous system. This disease is difficult to diagnose and is one of the common diseases in the public. In this paper, we have proposed a novel data sampling method for the classification of Parkinson disease based on the acoustic features from the speech signals. In the proposed data sampling method, the one against all (OGA) has been used to divide the dataset into five equal parts. With applying the OGA to the PD dataset having two classes (healthy and Parkinson disease), the minority and majority classes have been obtained. First of all, for healthy class in the dataset (first case), five equal partitions have been composed and then for PD class in the dataset (second case), five equal partitions have been composed. To classify the these all data partitions, we have used three different classifiers including the weighted k-NN (nearest neighbor), Logistic Regression (LR), and support vector machine with medium Gaussian kernel function. In order to evaluate the performance of the proposed hybrid models (the combination of classifiers and OGA based data sampling), the classification accuracy, the confusion matrix, and area under the Receiver Operating Characteristic (ROC) curve (AUC) have been used. While the LR, SVM with Gaussian, and weighted k-NN classifiers achieved the classification accuracies of 77.50%, 83.80%, and 82.10% in the classification of PD with the acoustic features, the combinations of classifiers and OGA based data sampling (first case) obtained the 79.04%, 87.36%, and 88.48% using the LR, SVM with Gaussian, and weighted k-NN classifiers, respectively. In the second case, the obtained classification accuracies are the 84.30%, 88.76%, and 89.46% using the LR, SVM with Gaussian, and weighted k-NN classifiers with the OGA based data sampling, respectively. The achieved results have shown that the proposed the one against all (OGA) based data sampling could be used in the combination of classifier algorithms as the data pre-processing method in the classification of Parkinson's disease with acoustic features.
      pubtype: Academic Journal
      doctype: Journal Article
      ougenre: Article
    language: English
    refInfo:
    holdings:
      @attributes:
        islocal: N