Hybrid Interpretable Deep Learning Framework for Skin Cancer Diagnosis: Integrating Radial Basis Function Networks with Explainable AI

AI can explain its medical decisions: Unlike typical "black box" AI systems, this framework uses Radial Basis Function Networks to show clinicians exactly which visual patterns in skin images led to a diagnosis, making the AI's reasoning transparent and trustworthy for clinical use.

Hybrid approaches may improve both accuracy and transparency: The combination of CNNs with RBF Networks achieved 83% accuracy on skin cancer classification while maintaining interpretability—demonstrating that AI systems don't have to choose between being accurate and being explainable.

Interpretability enables clinical oversight: By mapping predictions to specific, identifiable image features and prototypes, clinicians can verify whether the AI is basing decisions on medically relevant patterns, allowing them to catch potential errors or biases before they affect patient care.

Active learning focuses AI attention on what matters: The framework uses K-Medoids clustering and prototype selection to concentrate on salient features rather than processing entire images equally, which could reduce computational costs and help the system learn more efficiently from limited medical data.

Responsible AI in healthcare requires traceability: This work addresses a critical gap in medical AI—the need for systems where every diagnosis recommendation can be traced back to specific visual evidence, supporting both regulatory compliance and clinician confidence in automated diagnostic tools.