Blockchain-Integrated Support Vector Machine Framework for Privacy-Preserving Predictive Analytics

This paper presents a blockchain-integrated Support Vector Machine (SVM) framework that addresses the privacy, trust, and ownership limitations of conventional client–server architectures. In the proposed design, clients train local SVM models and share only encrypted parameters, while blockchain ensures secure aggregation, tamper-proof logging, and transparent auditability.This study proposes a framework that combines blockchain technology with Support Vector Machines (SVM) to improve privacy, transparency, and accountability in predictive analytics. Experiments on MNIST and CIFAR-10 show that the framework achieves accuracy comparable to conventional SVMs (0.92 vs. 0.93), while enabling tamper-resistant predictions and verifiable model operations. Blockchain integration introduces moderate overhead—latency of 18–19 ms and transaction costs around 22,500 gas units—but these are outweighed by gains in data security, decentralized control, and auditability. To the best of our knowledge, this is among the first efforts to merge SVM with blockchain for secure predictive modeling. The framework is scalable, reliable, and well-suited for sensitive domains such as healthcare, finance, and intelligent transportation systems.