Scalp diseases often manifest with overlapping clinical features and may co-occur within the same patient, posing significant diagnostic challenges in both clinical and remote set-tings. Existing AI-based tools for dermatological screening predominantly rely on single-label classification frameworks, which limit their applicability to real-world cases involving multiple concurrent conditions. In this study, we present an extension of the Skinner platform, an AI-driven mobile application for scalp disease detection, by introducing a multi-label classification model coupled with explainable artificial intelligence (XAI) mechanisms. The proposed architecture leverages a pre-trained ResNet-50 backbone and a custom multilayer perceptron classifier with sigmoid activation to independently estimate the presence of thirteen scalp disorders, including rare conditions such as trichotillomania and scalp lupus erythematosus. Class imbalance is addressed through data augmentation and class-weighted loss functions, while transparency is ensured through the integration of SHapley Additive exPlanations (SHAP), adapted for per-label visual interpretation. The system was evaluated on an extended, clinically curated dataset and demonstrated strong performance, achieving a macro F1-score of 0.791 and a mean average precision of 0.817. Qualitative analyses confirmed the medical plausibility of the model’s outputs and explanations. These findings underscore the potential of multi-label, explainable AI systems in enhancing diagnostic coverage, supporting clinical decision-making, and promoting user trust in automated dermatological screening tools.