International Research Journal of Education and Technology

Lung infections such as tuberculosis, pneumonia, COVID-19, and non-tuberculous mycobacterial (NTM) diseases remain a major global health concern, requiring timely and accurate diagnosis for effective treatment. Traditional diagnostic approaches often depend solely on either clinical parameters or radiological imaging, which may lead to incomplete assessments. This study presents an advanced AI-driven multimodal framework that combines clinical data with deep learning-based feature extraction from chest X-ray images. InceptionV3 and ShuffleNet architectures are utilized to capture essential visual features, which are then integrated with clinical parameters to create a comprehensive diagnostic model. The fused feature representation is processed through a classification model to improve predictive performance. Experimental results demonstrate that the proposed multimodal approach surpasses single-source methods in accuracy, sensitivity, and specificity. The findings highlight the effectiveness of integrating multiple data modalities for improved diagnostic reliability, offering a promising direction for AI-assisted lung infection detection in clinical practice.

Keywords—Lung infection classification, multimodal AI, deep learning, machine learning, chest X-ray analysis, clinical data integration, InceptionV3, ShuffleNet, XGBoost, early feature fusion.