Abstract

Chronic Obstructive Pulmonary Disease (COPD) remains critically under-diagnosed globally, contributing disproportionately to global respiratory morbidity and mortality. Development in deep learning modals like multimodal data fusion and Explainable Artificial Intelligence (XAI) automates COPD diagnosis. This literature review uses PRISMA to systematically examine the COPD diagnosis studies from 2007 to 2026, which were collected from Scopus and PubMed. Final 84 studies were analyzed with modals includes CT and CXR-based deep learning models, multimodal integration strategies, and XAI techniques like Grad-CAM, SHAP and LIME. The result shows that CT-based systems dominate the literature, whereas CXR, Spirometry, and clinical data were used in very limited studies. Explainable artificial intelligence (XAI) and clinical informatics are absent in the majority of the studies. This review identifies major research gaps and proposes a Hybrid Multimodal Explainable AI (HMX-COPD) framework by integrating Chest X-rays, Spirometry, and clinical data, providing scalability within a single interpretable architecture for COPD diagnosis.

Authors

N. Malathy, V. Vijaya Samundeeswari
Women’s Christian College, India

Keywords

COPD Diagnosis, Explainable AI, Deep Learning, SHAP, LIME