Classification model for Covid-19 detection through recording of cough using XGboost classifier algorithm

Irawati, M. E.; Zakaria, H.

Covid-19 is a worldwide outbreak of respiratory syndrome. To date, 178 million people have been exposed to Covid-19 since it was declared a global pandemic at the end of March 2020. Early detection of viral infections is critical in dealing with the outbreak's rapid spread. In this study, a machine learning model was developed that could classify Covid-19 based on cough sounds recorded. The MFCC feature extraction method is used to extract cough sound features from each cough recording, and these features are then used as learning input for machines to label using the XGBoost Classifier algorithm. With evaluation using cross validation method, the classification model's accuracy has reached 86.2%, allowing the model to be developed into a Covid-19 pre-screening tool that can reach a larger community. © 2021 IEEE.