Q-SHAP: Explainable AI for Quantum Circuits via Quantum-Native and Hybrid Classical Approaches

Quantum circuits are increasingly complex black-box systems that require interpretability for debugging, optimization, and validation. We present Q-SHAP, a comprehensive framework for explainable AI (XAI) in quantum computing, introducing quantum-native methods analogous to SHAP and LIME for classical machine learning. Our framework provides gate importance analysis using state fidelity metrics, Shapley value attribution for quantum gates, and entanglement flow tracking. We evaluate our approach on 10 benchmark quantum circuits including Bell states, GHZ states, Grover's algorithm, and variational quantum eigensolver (VQE) ansätze. Results show that quantum-native XAI methods achieve 0.84-0.98 consistency scores and successfully identify critical gates in quantum algorithms. We further compare quantum XAI with classical XAI applied to quantum-classical hybrid models, demonstrating that direct quantum state analysis provides superior interpretability for small-to-medium circuits (2-4 qubits). Our framework reduces circuit depth by up to 15% while maintaining >98% fidelity through intelligent gate pruning. This work enables practitioners to debug quantum algorithms, optimize gate sequences, and validate quantum advantage claims through interpretable explanations.