A Novel Approach to Multi-class Atopic Dermatitis Disease Severity Scoring using Multi-class SVM

The severity scoring of atopic dermatitis skin disease is a crucial stage for a dermatologist. Accurate and reliable decision-making by a dermatologist are needed for better therapy and risk stratification to the patient. Generally, the dermatologist used visual assessment to make decisions of the severity score. The procedure is very subjective and carries a lack of reliability. The paper presents a novel approach to classify the atopic dermatitis severity scoring based on machine learning algorithms using color, texture, and redness features of the skin. The redness features showed significant indication compared to other features. The overall accuracy achieved by the Multi-class SVM classifier was around 86%, and the multiclass SVM successfully classified the severity score into four classes.