Development and Performance Enhancement of Machine Learning based Approaches for Detection of Skin Diseases: A Survey

Skin diseases represent a major health challenge, and early detection is essential for successful treatment. Machine learning (ML) has emerged as a valuable tool in enhancing diagnostic accuracy. This study explores the application of Machine Learning for skin disease classification, emphasizing both image processing and clinical feature analysis. This article reviews the techniques and methods used to diagnose common skin conditions. Additionally, it has covered the assessment metrics for the performance analysis of different diagnosis systems as well as the image databases that are available. While feature based approaches consistently achieve accuracy above 94%, image-based methods show varied results ranging from 50% to 100%. Despite progress, issues such as data quality and model generalization persist. This review identifies key areas for further investigation and highlights the potential of Machine Learning to transform dermatological diagnostics. With the wide array of clinical and histopathological patterns observed in Papulosquamous skin disorder, precise differentiation is critical for accurate diagnosis and the development of effective, tailored treatment plans. The analysis provides an insightful overview of the relevant studies from the literature, highlighting the critical research gaps that have been identified to detect single skin lesion and imbalance dataset.