A Logical Evaluation of Tuberculosis Disease Prediction using Artificial Intelligence (AI) Enabled Learning Methodology

Tuberculosis (TB) remains a global health threat, requiring early and accurate detection to curb its spread. This study proposes a Hybrid Learning Model that integrates Transformer Networks, Capsule Networks, and LightGBM to predict tuberculosis using chest X-ray images and clinical data. The model leverages Transformer Networks to capture long-range dependencies in image data and Capsule Networks to capture spatial hierarchies, while LightGBM processes clinical data for final classification. Publicly available datasets, including the Montgomery County X-ray Set and Shenzhen X- ray Set, were used for model evaluation. The proposed model was compared against nine existing models, including traditional machine learning models like Logistic Regression and Random Forest, and deep learning models such as CNN, ResNet50, and DenseNet. The proposed hybrid model achieved an accuracy of 95.35%, outperforming all other models. These results highlight the efficacy of combining advanced image processing techniques with efficient classification algorithms for TB diagnosis.