A CDSS for Conducing Prediction and Prescription of Non-Cancerous Skin Disease

The rapid increase in skin diseases attribute to numerous factors, that includes environmental conditions, food habits, genetics and lifestyle changes which if ignored, causes diseases such as chronic dermatitis and psoriasis affecting human health. Prevention of skin diseases is best done when they are detected in their initial stages. The drought in research works on diagnosing skin diseases such as lichen planus and seborrheic dermatitis remains challenging due to overlapping clinical features and limitations in detecting subtle differences in the disease is identified through this research. Through our novel research work, a dataset of 4104 skin images of 4 disease classes including lichen planus, chronic dermatitis, psoriasis, and seborrheic dermatitis is collected from different public repositories and the prediction and prescription of skin diseases are carried out in 3 phases that include feature engineering in phase 1, the ResNet50 deep learning model is used to detect the skin disease class, CNN and VGG16 being used as reference models to justify the robust performance of the proposed ResNet50 model for accurate disease detection in phase 2. In addition, a Decision Support System (DSS) is designed to assist in diagnostic support for users to prove that they can provide prediction and remedial prescriptions using the Django framework.