AI-Based Skin Disease Detection and Treatment Recommendation: AI Dermacare
Mr. Arokiaraj Christian St Hubert 1 K. Kavya 2 E. Abinaya 3 K. Aishwarya 4
1 Assistant Professor, Department of Computer Science and Engineering, Sri Manakula Vinayagar Engineering College, Puducherry, India.
2 3 4 UG Student, Department of Computer Science and Engineering, Sri Manakula Vinayagar Engineering College, Puducherry, India.
ABSTRACT
Skin diseases are among the most common health issues worldwide, varying from mild conditions to severe chronic disorders that require timely diagnosis and treatment. Conventional skin disease prediction systems often utilize lightweight deep learning models such as MobileNet due to their efficiency and low computational requirements; however, these approaches may not achieve sufficient accuracy when handling complex and diverse dermatological datasets, leading to unreliable predictions in real-world applications. To address these limitations, the proposed system introduces HEPA-Net (H-E-P-A: Histogram Equalization, EfficientNetB0, Preprocessing, and Augmentation), a hybrid framework designed to enhance overall model performance. In this approach, histogram equalization improves image contrast, preprocessing ensures data consistency by removing noise and standardizing inputs, and augmentation increases dataset diversity to reduce overfitting, while EfficientNetB0 performs accurate feature extraction and classification. This integrated pipeline enables the system to capture fine-grained skin features more effectively, thereby improving classification accuracy while maintaining computational efficiency. The system further incorporates Explainable Artificial Intelligence (XAI) techniques to enhance transparency and user trust by generating visual explanations such as heatmaps, which assist users and healthcare professionals in understanding and validating predictions. Additionally, the system is designed to be user-friendly, allowing easy image input and quick interpretation of results, thereby supporting early detection and awareness of skin diseases. It also provides basic home remedy suggestions for minor conditions and recommends nearby dermatologists for further consultation, ensuring a complete healthcare support system. To ensure data privacy and security, patient information is encrypted and stored using the InterPlanetary File System (IPFS), providing a decentralized, secure, and tamper-resistant storage solution. Overall, the proposed system delivers an accurate, interpretable, secure, and user-centric solution for intelligent skin disease prediction and preliminary healthcare assistance.
Keywords: Skin Disease Prediction, EfficientNetB0, Explainable AI (XAI), IPFS, Deep Learning, Medical Image Analysis, Data Security.
