Computer Assisted Model for Auto Immune Skin Disease

A timely and accurate diagnosis of auto-immune skin diseases is essential for treatment. This project describes a new system involving deep learning techniques that detect auto-immune skin diseases. This is achieved via Convolutional Neural Networks (CNNs) that can categorize and identify skin diseases in real-time, assisting both patients and healthcare professionals. The project has data collection, preprocessing, model training and evaluation. The system is trained with quality medical datasets, including patient demographic data and skin image data, to ensure the system is robust and generalizable. This includes a trained model achieving an overall accuracy of 84% on a test dataset of 21,000 images. The development of a simple-to-use web interface with Flack that will allow for the reporting of the potential skin disease(s) in real-time to both you and the trained model. The user can upload images and receive diagnostic results that will empower them in their skin health decision making, including whether they need medical guidance in a timely manner. The system provides: Improved Diagnostic Accuracy: Reduces the chance for misdiagnosis, where the need for speedy and accurate treatment is crucial; Improved Access to Healthcare: The system provides diagnostic services to regions underserved by dermatologists; Improvement to Dermatologists: Provides accurate diagnostics used during consultations in developing personalized treatment pathways. Through an early detection and accurate diagnosis based on deep learning procedures, the project will help to improve the management of auto-immune skin diseases, which ultimately has an effect on improved treatment pathways and quality of life for patients. The principles of combining care with technology, one chip at a time.