Integrating Ensemble Learning into Remote Health Monitoring for Accurate Prediction of Oral and Maxillofacial Diseases

Due to excessive use of tobacco, oral and maxillofacial diseases are prevalent in Pakistan. This paper presents a deep learning-based approach for the accurate diagnosis of oral diseases, specifically focusing on mouth ulcers, hypodontia, and dental caries, using RGB images. Unlike previous studies that primarily utilize X-ray images, this research uses a diverse dataset of over 6,000 annotated RGB images. The methodology involves training and evaluating three models including VGG16, MobileNet, and InceptionV3 for individual disease classification. The models achieve high validation accuracies ranging from 90% to 95%. The weighted ensemble model, combining the predictions of the three models, is also implemented which resulted in an improved accuracy of 97%. The proposed methodology demonstrates the potential of deep learning in enhancing the precision and effectiveness of oral disease diagnosis, enabling timely intervention, and optimizing patient care. Future work could focus on expanding the dataset size to further improve the model's accuracy.