Skin cancer remains a global health concern, requiring early detection for improved patient outcomes. This paper presents a hybrid machine learning system that combines Convolutional Neural Networks (CNNs) for skin lesion image classification and CatBoost for demographic data analysis. Using a multimodal approach, the system processes high-resolution skin images and patient information to deliver accurate predictions. The ConvMixer architecture extracts robust visual features, while CatBoost leverages tabular data to predict disease categories. Ensemble integration further refines the output, achieving significant improvements in accuracy. Experimental results demonstrate the system’s effectiveness, making it a promising tool for dermatological diagnostics. Future works can include integration of Explainable AI (XAI) techniques, such as SHAP and Grad-CAM, to improve interpretability and optimizing the system for edge computing will enable real-time deployment on mobile devices.