DIAGNOSING PARKINSON'S DISEASE USING THE CLASSIFICATION OF SPEECH SIGNALS

This paper addressees the problem of an early diagnosis of Parkinson’s disease by the classification of characteristic features of person’s voice. A new, two-step classification approach is proposed. In the first step, the voice samples are classified using standard state-of-the-art classifiers. In the second step, the classified samples are assigned to patients and the final classification process based on majority criterion is performed. The advantage of using our new approach is the resulting, reliable patientoriented medical diagnose. The proposed two-step method of classification allows also to deal with the variable number of voice samples gathered for every patient. Preliminary experiments revealed quite satisfactory classification accuracy obtained during the performed leave-one-out cross validation. In this paper an important problem [5], [8] of the early diagnosis of Parkinson’s disease is addressed. The name of the disease comes from James Parkinson who described it in work [11]. Parkinson’s disease is a disorder of central nervous system and leads to many health issues, e.g., rigidity, imbalances, difficulty in talking, or slowness of movements. It is estimated that seven to ten million people currently suffer Parkinson’s disease worldwide. First symptoms are noted for people over the age of fifty (the average age of the onset of disease is about fifty nine). Parkinson’s disease manifests also in form of disorders of person’s speech. Therefore it is possible to diagnose Parkinson’s disease using voice signals [12], [14]. On the basis of voice sample a vector of numerical values is calculated. The obtained values represent selected characteristic features of the recorded voice. Finally, the characteristic vector of the voice sample is classified indicating whether the corresponding person exhibits a symptom of the disease. The problem of an early diagnosis of Parkinson’s disease has raised an interest of numerous researchers [3], [7], [13]. In particular, the application of artificial neural networks to discriminate healthy people from those with Parkinson’s disease using voice signals was proposed in [3]. Parkinson’s disease classification using gait characteristics and wavelet-based feature extraction was proposed in [13]. Application of fuzzy k-nearest neighbor model to an efficient detection of Parkinson’s disease was proposed in [7].