Eczema diagnosis and advisory system

Abstract

The Eczema Diagnosis and Advisory System (EDAS) is a tele-health platform designed to facilitate early detection, diagnosis and management of eczema, addressing the global challenge of limited access to dermatological care. This web-based system leverages machine learning and modern web technologies to empower patients and healthcare providers. EDAS allows users to upload skin images and complete a symptom-based questionnaire, the image is processed by a VGG19-based machine learning model to deliver diagnostic insights, including eczema detection, severity assessment, and confidence scores. The system provides treatment recommendations and lifestyle advice to promote self-care, while enabling appointment scheduling and direct communication with doctors for further consultation. It is built on a three-tier architecture, the frontend utilizes Next.js and TypeScript for a responsive user interface, while the backend, powered by Node.js and Express, employs a dual-database system (MySQL and MongoDB) for efficient data management. The machine learning service, developed with Flask and TensorFlow, ensures accurate image analysis and diagnostic reliability. Security features, including HIPAA and GDPR compliance, end-to-end encryption, and role-based access control, safeguard sensitive medical data. The system also offers analytics dashboards for tracking diagnoses, model performance, and severity distribution, enhancing decision-making for both patients and providers. Through rigorous testing, including unit, integration, and system acceptance tests, EDAS demonstrated high accuracy and reliability in supplementing clinical diagnoses. Conducted in collaboration with Mulago National Referral Hospital’s Skin Clinic, the project validated its potential to reduce diagnostic delays and costs, particularly in resource-constrained settings like Uganda. Limitations of this system include dependence on internet connectivity, mobile device requirements, and reliance on image-based diagnosis, which may lack comprehensive clinical context. Future enhancements aim to improve model accuracy, expand to other dermatological conditions and enhance global accessibility through multilingual support. EDAS represents a significant step toward easier dermatological care, offering a scalable, user-centric solution that bridges gaps in healthcare access while fostering patient empowerment and informed self-management.