MediCynth: An AI-Based Skin Disease Classification System

Abstract

With the rapid growth of computer technology, accurate and faster results can be obtained, allowing patients to receive timely treatment. Skin diseases are among the most common health problems and may be caused by viruses, bacteria, allergies, fungal infections, and other factors. Early and correct identification of skin diseases is important to prevent severe complications. This paper presents MediCynth, an automated skin disease detection system based on image processing and deep learning techniques. The proposed system analyzes an uploaded image of the affected skin area and predicts the type of skin disease. The model is trained using a combined dataset collected from DermNet, Kaggle, and additional internet sources. It is capable of classifying 27 different skin disease categories, including Acne, Lupus, Psoriasis, Skin Cancer, Vitiligo, Celluli-tis/Impetigo, Hair Loss/Alopecia, Herpes/HPV, Nail Fungus, and Urticaria/Hives. The system uses a Vision Transformer (ViT-B/16) architecture instead of traditional Convolutional Neural Networks (CNN), leveraging transfer learning to improve predic-tion accuracy and reduce diagnosis time. By processing a digital image of the skin, the proposed method identifies the disease and provides a confidence score along with basic guidance. The system is simple, fast, low-cost, and can be accessed from any device with an internet connection. Therefore, image processing and advanced deep learning techniques provide an effective solution for accurate skin disease detection and classification.