Published Paper Details:

Dermatological disease classification presents significant challenges due to limited labeled datasets, diverse lesion patterns, and the need for scalable diagnostic tools. Existing Explainable AI (XAI) techniques focus on interpretability but do not comprehensively address data scarcity, generalizability, and optimization. This paper proposes a hybrid AI framework integrating self-supervised learning (SSL), meta-learning (MAML), neural architecture search (NAS), and diffusion models. SSL pretrains on unlabeled data, MAML fine-tunes for domain-specific tasks, NAS automates model design, and diffusion models generate synthetic images to mitigate class imbalance. Ensemble learning improves robustness, while lightweight optimization ensures scalability for clinical deployment. Validation across datasets demonstrates improved accuracy, precision, recall, and F1-score. This framework offers a comprehensive solution to dermatological diagnostics by enhancing adaptability, robustness, and clinical applicability.