Hybrid Meta-Stacking System for Enhanced Diagnosis of Chronic Pulmonary Illnesses

The classification of chest X-rays has been extensively explored, yet achieving high diagnostic accuracy remains challenging due to factors like class imbalance and limited data. In this study, we propose a novel stacking-based ensemble learning approach that leverages pre-trained models and transfer learning to address these challenges. Our framework integrates DenseNet201, EfficientNetV2S, and InceptionResNetV2 as base models, with a robust meta-model for combining predictions. To mitigate class imbalance and enhance performance, we employed data augmentation using the augmenter package. Our stacking model significantly outperformed traditional methods such as concatenation, averaging, weighted averaging, and hard voting, achieving 97.89% precision and 97.76% recall on the test set. These results underscore the potential of ensemble learning with pre-trained models to improve diagnostic accuracy in medical imaging, paving the way for more reliable and efficient healthcare solutions.