SkinSenseAI: AI-Driven Skin Disease Detection with Retrieval-Augmented Clinical Decision Support

Abstract

Skin diseases are a widespread health concern, and delayed diagnosis often occurs due to limited access to dermatologists or lack of awareness. While deep learning models can detect skin conditions from images, many operate as “black boxes” without clear explanations, and standalone Large Language Models (LLMs) may produce inaccurate medical information. This paper presents SkinSense AI, an intelligent skin health assistant that combines image-based disease detection with Retrieval-Augmented Generation (RAG) to deliver accurate and explainable results. A CNN model identifies possible conditions from skin images, while the RAG system retrieves trusted medical knowledge to generate clear, evidence-based explanations, prevention tips, and lifestyle guidance. The platform also includes a questionnaire-based chatbot and a severity scoring module for early screening and triage. Results show that integrating computer vision with RAG improves both diagnostic reliability and user understanding. SkinSense AI serves as a scalable early-awareness tool that supports informed decision-making and encourages timely professional consultation.

Ähnliche Arbeiten