Comparison of COVID-19 Prediction Performances of Normalization Methods on Cough Acoustics Sounds

Abstract

The disease called the new coronavirus (COVID19) is a new viral respiratory disease that first appeared on January 13, 2020 in Wuhan, China. Some of the symptoms of this disease are fever, cough, shortness of breath and difficulty in breathing. In more serious cases, death may occur as a result of infection. COVID19 emerged as a pandemic that affected the whole world in a little while. The most important issue in the fight against the epidemic is the early diagnosis and follow-up of COVID19 (+) patients. Therefore, in addition to the RT-PCR test, medical imaging methods are also used when identifying COVID 19 (+) patients. In this study, an alternative approach was proposed using cough data, one of the most prominent symptoms of COVID19 (+) patients. The performances of z-normalization and min-max normalization methods were investigated on these data. All features were obtained using discrete wavelet transform method. Support vector machines (SVM) was used as classifier algorithm. The highest performances of accuracy and F1-score were obtained as 100% and 100% using the min-max normalization, respectively. On the other hand, the highest accuracy and highest F1-score performances were obtained as 99.2 % and 99.0 % using the z-normalization, respectively. In light of the results, it is clear that cough acoustic data will contribute significantly to controlling COVID19 cases.