Abstract

The pervasive impact of human-to-human disease transmission on society is evident in recent global pandemic out-breaks. Identifying and managing such diseases pose significant challenges, exacerbated by the absence of specific medications due to the extensive variants involved. Their similarities to pneumonia heighten the complexity of distinguishing these diseases. To address this, emerging machine learning (ML) and deep learning (DL) models offer promising solutions. This study delves into the potential of DL models, surpassing traditional ML, for more meaningful disease classification. Leveraging well-established models and lung-related X-ray images (CXR), we employ a transfer learning-based approach with meticulous hyperparameter tuning. Our results demonstrate improved classification outcomes compared to previous attempts on the same dataset. Furthermore, our work highlights the optimization potential of existing modeling techniques, showcasing the feasibility of refining approaches with available resources for similar purposes.