MouthCare: AI Powered Mobile Application for Oral Disease Classification Using an Optimized DenseNet-121 Model

Access to dental care in the Philippines is limited due to high and inconsistent costs, creating a need for affordable solutions. This study presents “MouthCare,” an AI-powered mobile app for detecting cavities, calculus, and gingivitis using an optimized DenseNet-121 deep learning model. Through transfer learning, fine-tuning, and TensorFlow Lite, the model is efficiently deployed on mobile devices, overcoming challenges of overfitting and computational constraints. It achieved a 99.50% accuracy, a 99.97% AUC, and a 3.40-second latency, demonstrating its potential for real-time use. The model's implementation on Android devices offers a scalable approach to improving oral healthcare accessibility in the Philippines. Future work should expand the dataset to include more oral diseases, improving the model's generalizability and clinical impact.