Integrating Data Augmentation with AdaBoost for Effective COVID-19 Pneumonia Classification

Abstract

In the COVID-19 pandemic, the urgent need for precise pneumonia diagnosis has prompted this research to propose a customized AdaBoost algorithm tailored for classifying COVID-19 pneumonia cases. The study follows a structured framework encompassing four primary supervised learning stages: data acquisition, preprocessing, supervised learning with augmented data, and rigorous performance evaluation. Leveraging a comprehensive GitHub dataset and Python in Anaconda Jupyter Notebook, advanced preprocessing techniques are employed to optimize data for machine learning algorithms. The AdaBoost algorithm, enhanced with data augmentation methods, is deployed to bolster model robustness and performance. The model demonstrates superior effectiveness with an average accuracy rate of 84.49%, surpassing existing methodologies. This performance underscores its potential in addressing public health challenges associated with pneumonia diagnosis during the COVID-19 crisis. This research introduces an optimized application of AdaBoost for pneumonia classification, validated across diverse datasets, ensuring reliable disease classification and predictive modeling capabilities to anticipate future trends. These insights are pivotal for guiding public health interventions and optimizing resource allocation, marking significant advancements in diagnostic accuracy and patient care during the pandemic.