A Distributed Adaptive Differential Privacy Mechanism With Blockchain Integration for Edge Computing

ABSTRACT With the proliferation of big data‐driven healthcare, edge computing, and blockchain sharing, privacy protection for sensitive clinical data remains challenging. Conventional differential privacy (DP)—with fixed budgets, centralized noise injection, and no auditability—underperforms in distributed settings. We propose Distributed Adaptive Differential Privacy with Blockchain Integration (DADP‐BI): each edge node dynamically allocates budgets by data sensitivity/query frequency, applies Kalman smoothing to local Laplace noise, and uses smart contracts for on‐chain audit. Evaluated on the expanded UCI Heart Disease Dataset (920 samples, 14 attributes including high‐sensitivity “cholesterol” and “fastingBS”): Privacy leakage dropped by 49.3% (21.7% → 10.8% for “cholesterol”) and 41.4% (15.2% → 8.9% for “age”); Regression errors fell (MSE: 0.38 → 0.22, MAE: 0.27 → 0.16), and heart disease classification accuracy rose 7.2 percentage points (77.8% → 85.0%); End‐to‐end runtime and blockchain latency decreased by 18.5%–19.8%, with CPU/memory usage down 31.8%/26.7%. DADP‐BI offers a scalable solution for privacy‐preserving data processing in decentralized edge big‐data environments.