Detection of Parkinson's Disease using Speech Signals with L1-Norm SVM and Chi-Square Based Feature Selection Algorithms

Yıl 2021, , 32 - 40, 30.06.2021

Fatih Demir

https://doi.org/10.29132/ijpas.873653

Cited By: 1

Öz

Parkinson's disease is one of the common neurodegenerative disorders. Speech / voice impairment is considered as one of the early symptoms of Parkinson's disease. Sound signal processing methods can potentially assess and measure Parkinson's disease-related sound impairment. In this study, an effective machine learning technique has been proposed to diagnose Parkinson's disease from speech signals. In the proposed method, a data set containing features extracted from speech signals of healthy people and Parkinson's patients was used. Highly distinctive features in the data set were selected using L1-Norm Support Vector Machine and Chi-Square Based feature selection algorithms. The feature sets obtained from the two methods were combined and used in the classification stage. In the classification stage, the achievement of proposed method was increased with majority voting method, which used the prediction results of Support Vector Machine, K-Nearest Neighbor and Random Subspace K-Nearest Neighbor Ensembles classifiers. The proposed method with 95.11% accuracy outperformed previous studies using the same dataset. Since Parkinson's disease will be diagnosed automatically with the proposed method, this application can be used as a helpful tool for physicians in their decision-making process.

Anahtar Kelimeler

Classification, feature selection, Parkinson’s disease, speech signal

L1-Norm DVM ve Ki-Kare Tabanlı Öznitelik Seçme Algoritmaları ile Parkinson Hastalığının Konuşma Sinyalleri Üzerinden Saptanması

Öz

Parkinson hastalığı, genel nöro-dejeneratif bozukluklardan biridir. Konuşma / ses bozukluğu Parkinson hastalığının erken dönemdeki semptomlarından biri olarak kabul edilir. Ses sinyallerini işleme yöntemleri, Parkinson hastalığı ile ilgili ses bozukluğunu potansiyel olarak değerlendirebilir ve ölçebilir. Bu çalışmada, Parkinson hastalığını konuşma sinyallerinden teşhis etmek için etkili bir makine öğrenmesi tekniği önerilmiştir. Önerilen yöntemde, sağlıklı kişilerin ve Parkinson hastalarının konuşma sinyallerinden çıkarılan öznitelikleri içeren bir veri seti kullanılmıştır. L1-Norm Destek Vektör Makinesi ve Ki-Kare Tabanlı öznitelik seçme algoritmaları kullanılarak veri setinde bulunan ayırt ediciliği yüksek öznitelikler seçilmiştir. İki yöntemden elde edilen öznitelik setleri birleştirilerek sınıflandırma aşamasında kullanılmıştır. Sınıflandırma aşamasında Destek Vektör Makinesi, K-En Yakın Komşu ve Rasgele Alt Uzay K-En Yakın Komşu Toplulukları sınıflandırıcılarının tahmin sonuçlarının kullanıldığı Çoğunluk Oylaması yöntemi ile önerilen yöntemin başarımı artırılmıştır. Önerilen yöntem %95.11 doğruluk ile aynı veri setini kullanan geçmiş çalışmalara göre daha iyi bir performans sağlamıştır. Önerilen yöntem ile Parkinson hastalığı otomatik olarak teşhis edileceğinden bu uygulama hekimlere karar verme süresinde yardımcı bir araç olarak kullanılabilir.

Anahtar Kelimeler

Konuşma sinyalleri, öznitelik seçme, Parkinson hastalığı, sınıflandırma

Kaynakça

• P. Calabresi, B. Picconi, L. Parnetti, and M. Di Filippo, “A convergent model for cognitive dysfunctions in Parkinson’s disease: the critical dopamine--acetylcholine synaptic balance,” Lancet Neurol., vol. 5, no. 11, pp. 974–983, 2006.
• J. Jankovic, “Parkinson’s disease: clinical features and diagnosis,” J. Neurol. Neurosurg. psychiatry, vol. 79, no. 4, pp. 368–376, 2008.
• A. H. Friedlander, M. Mahler, K. M. Norman, and R. L. Ettinger, “Parkinson disease: systemic and orofacial manifestations, medical and dental management,” J. Am. Dent. Assoc., vol. 140, no. 6, pp. 658–669, 2009.
• B. R. Bloem, Y. A. M. Grimbergen, M. Cramer, M. Willemsen, and A. H. Zwinderman, “Prospective assessment of falls in Parkinson’s disease,” J. Neurol., vol. 248, no. 11, pp. 950–958, 2001.
• A. S. Ashour, A. El-Attar, N. Dey, M. M. Abd El-Naby, and H. Abd El-Kader, “Patient-dependent freezing of gait detection using signals from multi-accelerometer sensors in Parkinson’s disease,” in 2018 9th Cairo International Biomedical Engineering Conference (CIBEC), 2018, pp. 171–174.
• M. Trail, C. Fox, L. O. Ramig, S. Sapir, J. Howard, and E. C. Lai, “Speech treatment for Parkinson’s disease,” NeuroRehabilitation, vol. 20, no. 3, pp. 205–221, 2005.
• L. O. Ramig, C. Fox, and S. Sapir, “Parkinson’s disease: speech and voice disorders and their treatment with the Lee Silverman Voice Treatment,” in Seminars in speech and language, 2004, vol. 25, no. 02, pp. 169–180.
• A. M. Goberman, “Correlation between acoustic speech characteristics and non-speech motor performance in Parkinson disease,” Med. Sci. Monit., vol. 11, no. 3, pp. CR109--CR116, 2005.
• J. R. Orozco-Arroyave et al., “Automatic detection of Parkinson’s disease in running speech spoken in three different languages,” J. Acoust. Soc. Am., vol. 139, no. 1, pp. 481–500, 2016.
• A. Tsanas, M. A. Little, P. E. McSharry, J. Spielman, and L. O. Ramig, “Novel speech signal processing algorithms for high-accuracy classification of Parkinson’s disease,” IEEE Trans. Biomed. Eng., vol. 59, no. 5, pp. 1264–1271, 2012.
• C. O. Sakar et al., “A comparative analysis of speech signal processing algorithms for Parkinson’s disease classification and the use of the tunable Q-factor wavelet transform,” Appl. Soft Comput., vol. 74, pp. 255–263, 2019.
• K. Kira and L. A. Rendell, “A practical approach to feature selection,” in Machine learning proceedings 1992, Elsevier, 1992, pp. 249–256.
• A. U. Haq et al., “Feature selection based on L1-norm support vector machine and effective recognition system for Parkinson’s disease using voice recordings,” IEEE access, vol. 7, pp. 37718–37734, 2019.
• P. S. Bradley and O. L. Mangasarian, “Feature selection via concave minimization and support vector machines.,” in ICML, 1998, vol. 98, pp. 82–90.
• S. Guo, D. Guo, L. Chen, and Q. Jiang, “A L1-regularized feature selection method for local dimension reduction on microarray data,” Comput. Biol. Chem., vol. 67, pp. 92–101, 2017.
• C. Cortes and V. Vapnik, “Support vector machine,” Mach. Learn., vol. 20, no. 3, pp. 273–297, 1995.
• M. Turhan, D. Sengür, S. Karabatak, Y. Guo, and F. Smarandache, “Neutrosophic weighted support vector machines for the determination of school administrators who attended an action learning course based on their conflict-handling styles,” Symmetry (Basel)., vol. 10, no. 5, p. 176, 2018.
• D. Sengür and M. Turhan, “Prediction of the action identification levels of teachers based on organizational commitment and job satisfaction by using k-nearest neighbors method,” Turkish J. Sci. Technol., vol. 13, no. 2, pp. 61–68, 2018.
• T. Cover and P. Hart, “Nearest neighbor pattern classification,” IEEE Trans. Inf. theory, vol. 13, no. 1, pp. 21–27, 1967.
• K. Adem, “Diagnosis of breast cancer with Stacked autoencoder and Subspace kNN,” Phys. A Stat. Mech. its Appl., vol. 551, p. 124591, 2020.
• T. K. Ho, “Nearest neighbors in random subspaces,” in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 1998, vol. 1451, pp. 640–648, doi: 10.1007/bfb0033288.
• A. S. Ashour, M. K. A. Nour, K. Polat, Y. Guo, W. Alsaggaf, and A. El-Attar, “A Novel Framework of Two Successive Feature Selection Levels Using Weight-Based Procedure for Voice-Loss Detection in Parkinson’s Disease,” IEEE Access, vol. 8, pp. 76193–76203, 2020.