אפקה - פרויקטי גמר 2020

Diagnosis of Parkinson’s Disease in Human Using

Voice Signals

Background

•

Parkinson's disease (PD) is a chronic and progressive disorder that affect the

nervous system that usually affects people over the age of 60.

•

Speech disorder in Parkinson (also known as part of hypokinetic Dysarthria

(

,the

main cause of the disorder is muscle disorders in the speech mechanism due to

damage of the central nervous system.the main problems that can be noticed

happens due to paralysis, weakness, or disobedience of speech muscles, The

disorder increases with time as the disease progresses and can have tangible

effect on communication skills.

Student name: Yazid Bisharat

Advisor: Yirmi Hauptan

Electrical Engineering

Classifier

The feature that have been trained was SVM with three different kernels, and with three

parameters

building classifier for diagnosis of Parkinson’s Disease in Human

Using Voice Signals, that can easily be used and trying to detect

patients.

Objective

Train classifier that can detect

Parkinson's disease speech disorder with accuracy

more than 80%

method

The speech task that were used in this project was reading text task, The recordings

were in .WAV format with 4410 sampling frequency, the speech tasks that I used

from the records is the reading text tasks, the recordings consists of Healthy controls

68( 42 men, 42 women) with age mean of 63.04 std1.572, and Parkinson patients

(39 men, 27 women) and PD with age mean 64.39 std 10.4

feature selection:

22 acoustic features like HNR, shimmer and jitter have been

extracted from speech recordings.

Features have significant difference

SVM with 4 degree Polynomial kernel

Accuracy =

0.805 (+/- 0.088)

Results

SVM with RBF kernel

Accuracy =

.818 (+/- 0.146)

SVM with linear kernel

Accuracy = 0.805 (+/- 0.088)

Conclusion

The lassfier that gave the best re