RECOGNITION AND ANALYSIS OF ASTHMA DISEASE BASED ON COUGH SOUND
Naveen Kumar Vodnala1, Y. Padma Sai2, Prathyusha Vilasagaram3, *
1 Assistant Professor, Department of Electronics and Communication Engineering, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad, India
2 Professor, Department of Electronics and Communication Engineering, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad, India
3 M.Tech student, Department of Electronics and Communication Engineering, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad, India
* Corresponding author-E-mail: vilasagaram.prathyusha@gmail.com
Abstract
Objective: To identify asthma based on cough sounds as the cough is the primary symptom that occurs when the lung airways are being deteriorated.
Methods: The cough sound signals are decomposed using the Discrete Wavelet Transform to obtain wavelet coefficients which are used in extracting some spectral and time-domain features. The extracted features are given as input to machine learning-based classifiers that are Decision Tree, Support vector machine and K-Nearest Neighbour to discriminate between Cough sound signals.
Findings: The maximum accuracy of 92.3% is obtained from the KNN classifier. The specificity and sensitivity obtained are 100% and 85% respectively. F1-score obtained is also greater for the KNN classifier comparatively.
Novelty: This paper includes the Classification of Asthma cough sounds from Normal cough sounds with the combination of wavelet decomposition with comparative of the three different types of classifiers and also computes their performance states the novelty of this research article.
Keywords: Asthma, K-Nearest Neighbour, Raspberry Pi, cough, Support vector machine, Discrete Wavelet Transform, Decision Tree.