Explainable AI-Enabled Privacy-Preserving Query Processing on Blockchain Ledgers With Statistical Metadata

Blockchain has gained increasing attention for managing eHealth data, yet most existing ledger designs face fundamental challenges in simultaneously enforcing strict privacy protections, enabling low-latency data queries, and supporting transparent AI-driven decision-making. This paper proposes a novel blockchain ledger architecture integrating Explainable Artificial Intelligence (XAI) through SHapley Additive exPlanations (SHAP) and statistical metadata for privacy-preserving and efficient query processing in eHealth applications. Our method introduces dynamic, expert-informed sensitivity classification and interpretable SHAP values directly embedded in block headers. The integration significantly reduces query data requirements by approximately 99.78%, maintaining patient data privacy through role-based access control and metadata-driven querying. Empirical validation in an IoMT-enabled healthcare scenario confirms substantial query efficiency, preserving privacy and transparency improvements, justifying the computational overhead associated with metadata creation. The results highlight the practical utility and robustness of our proposed blockchain architecture for rapid clinical decision-making.