View article

Asthma is among the most common condition, and it has been reported that it is currently poorly controlled. In this work, wheeze sound were analysed to classify different levels of asthma severity. Wheeze signals were obtained from patients with three asthma severity levels, namely mild, moderate and severe asthma. The wheeze sounds detected were then used for a feature extraction process using mel-frequency cepstral coefficients (MFCC). The extracted features were then evaluated by one-way ANOVA, and the MFCC features were then subjected to a classification process using the K-nearest neighbour (KNN) algorithm for the classification process. The performance of the KNN was found to exhibit an average accuracy of 97.5%. This study reveals that wheeze analysis is a competent approach for designing a computerized system for monitoring severity of asthma based on wheeze sounds.