Doğrusal Olmayan Analiz ve Makine Öğrenimi Kullanılarak Öksürük Seslerinden COVID-19'un Tanımlanması

Abstract

COVID-19'un otomatik teşhisi, insanlarla etkileşimi en aza indirerek hastalığın yayılmasını azaltmada aktif bir role sahiptir. Çeşitli sinyal ve görüntüleri kullanan makine öğrenmesi modelleri, otomatik tanılamanın temelini oluşturur. Bu çalışma, COVID-19 ve COVID-19 değil olarak etiketlenmiş öksürük ses sinyallerini içeren 'Virufy' veri setini kullanarak COVID-19 enfeksiyonunu tespit etmek için makine öğrenmesi tabanlı modeller sunmaktadır. Veri setindeki COVID pozitif öksürük sayısı, COVID negatif olanlardan daha az olduğu için çalışmada öncelikle ADASYN aşırı örnekleme tekniği ile veri dengeleme yapılmıştır. Ardından, Çokfraktallı Eğimden Arındırılmış Dalgalanma Analizi (Multifraktal Detrended Fluctuation Analysis - MDFA), Lempel-Ziv Karmaşıklığı (Lempel–Ziv Complexity-LZC) ve entropi ölçümleri kullanılarak öksürük seslerinin doğrusal olmayan analizi ile öznitelikler çıkarılmıştır. Daha sonra ReliefF yöntemi ile en etkili öznitelikler seçilmiştir. Son olarak, öksürük seslerini COVID-19 veya değil olarak tanımlamak için, Radyal Tabanlı Çekirdek fonksiyona sahip Destek vektör Makineleri (Support Vector Machine with Radial Basis Function-SVM-RBF), Rastgele Orman (Random Forest-RF), Adaboost, Yapay Sinir Ağları (Artificial Neural Network -ANN), k En Yakın Komşuluk (k Nearest Neighbor -kNN) olmak üzere beş makine öğrenme algoritması kullanılmıştır. Çalışma sonucunda, radyal tabanlı çekirdek fonksiyonuna sahip destek vektör makinesi ve seçilen etkin öznitelikler sayesinde COVID-19 hastalarının ve COVID19 olmayan deneklerin öksürük sesleri %95.8 sınıflandırma doğruluğu ile belirlenmiştir. Bu sınıflandırıcı ile %93.1 duyarlılık, %98.6 özgüllük, %98.6 kesinlik, 0.92 kappa istatistik değerleri ve %93.2 ROC eğrisi altında kalan alan değeri elde edilmiştir.

Keywords

• COVID-19
• Öksürük Sesleri
• Entropi
• Makine Öğrenimi
• MDFA
• LZC
• SVM.

Identification of COVID-19 from Cough Sounds Using Non-Linear Analysis and Machine Learning

Abstract

Automatic diagnosis of COVID-19 has an active role in reducing the spread of the disease by minimizing interaction with people. Machine learning models using various signals and images form the basis of automatic diagnosis. This study presents the machine learning based models for detecting COVID-19 infection using ‘Virufy’ dataset containing cough sound signals labeled as COVID-19 and Non-COVID-19. Since the number of COVID positive coughs in the set is less than those of COVID negative, firstly, data balancing was performed with the ADASYN oversampling technique in the study. Then, features were extracted by non-linear analysis of cough sounds using Multifractal Detrended Fluctuation Analysis (MDFA), Lempel–Ziv Complexity (LZC) and entropy measures. Later, the most effective features were selected by ReliefF method. Finally, five machine learning algorithms, namely Support Vector Machine with Radial Basis Function (SVM-RBF), Random Forest (RF), Adaboost, Artificial Neural Network (ANN), k Nearest Neighbor (kNN) were used to identify cough sounds as COVID-19 or Non-COVID19. As a result of the study, the cough sounds of COVID-19 patients and Non-COVID19 subjects were identified with 95.8% classification accuracy thanks to the RBF kernel function of SVM and the selected effective features. With this classifier, 93.1% sensitivity, 98.6% specificity, 98.6% precision, 0.92 kappa statistical values and 93.2% area under the ROC curve were obtained.

Keywords

• COVID-19
• Cough Sounds
• Entropy
• Machine Learning
• MDFA
• LZC
• SVM.

References

1. A. Fakhry, X. Jiang, J. Xiao, G. Chaudhari, A. Han, and A.Khanzada, "Virufy: A Multi-Branch Deep Learning Network for Automated Detection of Covid-19," preprint from arXiv:2103.01806, 2021.
2. A. Imran, I. Posokhova, H. N. Qureshi, U. Masoos, M. S. Riaz, K. Ali,C. N. John, M. I. Hussain, and M. Nabeel, "AI4COVID-19: AI enabled preliminary diagnosis for COVID-19 from cough samples via an app.," Inform Med Unlocked, vol. 20, pp. 100378, 2020.
3. A. Lempel, and J. Ziv, “ On the Complexity of Finite Sequences,” IEEE Transactions on Information Theory, vol. 22, no.1, pp. 75-81,1976.
4. A. Mahmoud, K. H. Rahouma, and S.M. Ramzy, "Pay attention to the speech: COVID-19 diagnosis using machine learning and crowdsourced respiratory and speech recordings," Alexandria Engineering Journal, 2021. In press.
5. A. Narin, C. Kaya, and Z. Pamuk, "Automatic detection of coronavirus disease (COVID-19) using X-ray images and deep convolutional neural networks," Pattern Anal Appl, pp. 1-14, 2021.
6. A. Pal, and M. Sankarasubbu, "Pay Attention to the cough: Early Diagnosis of COVID-19 using Interpretable Symptoms Embeddings with Cough Sound Signal Processing," in 36th ACM/SIGAPP Symposium on Applied Computing (SAC ’21). March 22–26, 2021.
7. E. A. Ihlen, "Introduction to multifractal detrended fluctuation analysis in matlab," Front Physiol, vol. 3, pp. 141, 2012.
8. F. Kaspar, and H. Schuster, "Easily calculable measure for the complexity of spatiotemporal patterns," Phys Rev A Gen Phys, vol. 36, no. 2, pp. 842-848, 1987.

9. F. Z. Göğüş, G. Tezel, S. Özşen, S. Küççüktürk, H. Vatansev, and Y. Koca, "Identification of Apnea-Hypopnea Index Subgroups Based on Multifractal Detrended Fluctuation Analysis and Nasal Cannula Airflow Signals," Traitement du Signal, vol. 37, no. 2, pp. 145-156, 2020.
10. G. Chaudhari, X. Jiang, A. Fakhry, A. Han, j. Xiao, S. Shen, and A. Khanzada, "Virufy: Global Applicability of Crowdsourced and Clinical Datasets for AI Detection of COVID-19 from Cough," ArXiv, 2020. 2011.13320.
11. H. Coppock, A. Gaskell, P. Tzikaris, A. Baird, L. Jones, and B. W. Schuller, "End-2-End COVID-19 Detection from Breath & Cough Audio," Preprint from arXiv:2102.08359v1, 2021.
12. H. He, Y. Bai, E. A. Garcia, and S. Li, "ADASYN: Adaptive synthetic sampling approach for imbalanced learning," in 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence), pp. 1322-1328, 2008.
13. I. Kononenko, "Estimating Attributes: Analysis and Extensions of Relief," in European Conference on Machine Learning, pp. 171-182, 1994.
14. J. S. Richman, and J. R. Moorman, "Physiological time-series analysis using approximate entropy and sample entropy," Am J Physiol Heart Circ Physiol, vol. 278, no. 6, pp. H2039-49, 2000.
15. L. F. Márton, S. t. Brassai, L. Bako, and L. Losonczi, "Detrended Fluctuation Analysis of EEG Signals," Procedia Technology, vol. 12, pp. 125-132, 2014.
16. M. J. Horry, S. Chakraborty, M. Paul, A. Ulhaq, B. Pradhan, M. Saha, and N. Shukla, "COVID-19 Detection Through Transfer Learning Using Multimodal Imaging Data," IEEE Access, vol. 8, pp. 149808-149824, 2020.
17. M. N. Manshouri, "Identifying COVID-19 by using spectral analysis of cough recordings: a distinctive classification study," Cogn Neurodyn, pp. 1-15, 2021.
18. M. Pahar, M. Klopper, R. Warren, and T. Niesler, "COVID-19 cough classification using machine learning and global smartphone recordings," Comput Biol Med, vol. 135, pp. 104572, 2021.
19. P. Bagad, A. Dalmia, J. Doshi, A. Nagrani,P. Bhamare, A. Mahale, S. Rane, N. Agarwal, and R. Panicker, "Cough Against: COVID Evidence of COVID-19 Signature in Cough Sounds," Preprint from arXiv, 2020.
20. P. Mouawad, T. Dubnov, and S. Dubnov, "Robust Detection of COVID-19 in Cough Sounds: Using Recurrence Dynamics and Variable Markov Model," SN Comput Sci, vol. 2, no. 1, pp. 34, 2021.
21. S. M. Pincus, "Approximate entropy as a measure of system complexity," Proc. Nati. Acad. Sci., vol. 88, pp. 2297-2301, 1991.
22. Virufy COVID-19 Open Cough Dataset, GitHub - virufy/virufy_data, (n.d.). https://github.com/virufy/virufy_data (accessed October 08, 2021). Virufy, Editor. 2020.