EpiSure: AI-Powered Skin Disease Diagnosis Using MedGemma for Early Detection and Preventive Care

Abstract

Skin diseases are among the most common health conditions affecting millions of people worldwide and can significantly impact an individual’s health and quality of life. Early and accurate detection of skin diseases plays a crucial role in ensuring appropriate treatment and preventing disease progression. However, many existing diagnostic systems rely primarily on visual inspection or single-source data, such as images alone, which limits their accuracy and reliability. To address these challenges, the proposed system, EpiSure, leverages MedGemma, a multimodal generative AI model designed for medical data analysis. The system integrates multiple data sources, including skin images, clinical text records, and patient history, to enable more comprehensive and accurate diagnosis. By applying deep learning techniques and multimodal feature fusion, EpiSure enhances the consistency, reliability, and interpretability of skin disease prediction outcomes. The system architecture incorporates image preprocessing for noise reduction, feature extraction using a Vision Transformer model, and multimodal data integration within the MedGemma framework. Additionally, EpiSure provides healthcare professionals with an intuitive dashboard that presents the predicted disease type, severity level, and recommendations for early medical intervention.