An AI Based Multi-Modal System for Thyroid Disease Using Xgboost and Resnet18

Abstract

Thyroid disease are common disorders requiring timely diagnosis. This project presents a multi-modal AI system combining XGBoost for structured clinical data and ResNet18 for thyroid imaging to classify and predict thyroid disease accurately. The system integrates predictions using a fusion module and provides medicine suggestions, automated medical reports, and an AI chatbot to guide patients on whether to visit a hospital. Experimental results show improved accuracy over single-model approaches, offering a reliable, real-time and interpretable solution to assist clinicians in early detection, treatment planning, and better patient outcomes.