Detection of Parkinson Disease Using Frequency Sub-bands of Vocal Cords Vibration Signals and Machine Learning Techniques

Authors : Büşra Zeynep Gürel, Kübra Tancı, Mahmut Hekim, Cem Emeksiz

Pages : 101-113

View : 27 | Download : 39

Publication Date : 2024-11-30

Article Type : Research Paper

Abstract :In this study, we proposed a new approach for diagnosing Parkinson’s disease (PD) based on the slope values between neighboring amplitudes of vocal cord vibration signals. The inter-amplitude slope signals were obtained by computing the slopes between adjacent amplitudes in the vocal cord vibration signals. Feature vectors were extracted using common statistical parameters and applied to widely used machine learning classifiers such as Naive Bayes (NB), Generalized Logistic Regression (GLR), Logistic Regression (LR), Decision Tree (DT), and Random Forest (RFs). Different experiments were conducted to evaluate the contribution of the inter-amplitude slope approach and the performance of the classifiers in distinguishing healthy and PD segments. The experiments were carried out on original signals, inter-amplitude slope signals, and sub-band decompositions of both original and slope signals. The results showed satisfactory classification accuracy for all feature extraction methods, with the highest accuracy achieved using inter-amplitude slope signals. The GLR and Random Forest (RFs)-based classifiers outperformed others, achieving 100% accuracy, while the LR classifier reached 91%, and the DT and NB classifiers achieved 95%. Finally, the inter-amplitude slope approach, used for the first time in this study, enhanced classifier performance in PD diagnosis.
Keywords : Ses telleri, Parkinson Hastalığı, özellikler arası eğim