Artificial Intelligence Methods for Multiclass Identification of Skin Diseases

Survival in the treatment of skin diseases depends on early detection and accurate diagnosis. There is a need to provide accessible self-assessment solutions for the early recognition of pathological skin lesions. In treatment, early diagnosis is a critical factor for disease remission, effective mitigation of its impact, and improved survival chance. The proposed study focuses on using machine learning methods for analysis and identification of dermatological changes to provide an effective classification system. The dataset used was analysed depending on various factors including age, disease location and gender. In contrast to current solutions based only on two types of changes, the research enables multi-class recognition of pathological skin lesions. Based on the obtained results, it can be stated that the presented solution demonstrated the ability to accurately identify seven skin disease types, achieving an overall accuracy of 80%.