AI-Powered Skin Disease Detection Using Hybrid Learning Techniques

Early detection of skin diseases is crucial for effective treatment, yet accessibility to dermatologists remains limited. This research presents a conversational AI chatbot that integrates ResNet-18 for image classification and natural language processing (NLP) for symptom-based diagnosis. The system allows users to upload images or describe symptoms, providing a preliminary diagnosis and recommendations. A fine-tuned ResNet-18 model classifies skin disease images, while an NLP-based decision tree model processes textual descriptions. The chatbot is deployed using a Flask-based API, integrating Google Dialogflow for real-time interactions. Experimental results demonstrate a classification accuracy of 94.1%, validating the system’s capability as an assistive tool in dermatological assessments. Future enhancements include dataset expansion, improved model generalization, and integration with telemedicine services to enhance real-world applicability.