A Security Model for Real‑Time Fraud Detection and Prevention in Digital Payment Platforms Using Neural Networks and Blockchain
TL;DR
The findings indicate that neural architectures particularly recurrent networks, autoencoders, transformers, and graph neural networks provide superior adaptability to evolving fraud behaviour compared with static rule-based systems, but their operational deployment is constrained by class imbalance, concept drift, and regulatory demands for explainability.
Abstract
This review synthesizes recent research on real‑time fraud detection and prevention in digital payment platforms through the combined use of neural networks and blockchain technologies. A structured literature review and mapping study was conducted using PRISMA‑style screening procedures. Searches were performed across IEEE Xplore, ACM Digital Library, ScienceDirect, SpringerLink, and Google Scholar for studies published between 2015 and 2025. After applying inclusion/exclusion criteria and a quality appraisal checklist, forty (40) primary studies and ten (10) high‑impact survey papers were retained for analysis. The findings indicate that neural architectures particularly recurrent networks, autoencoders, transformers, and graph neural networks provide superior adaptability to evolving fraud behaviour compared with static rule‑based systems, but their operational deployment is constrained by class imbalance, concept drift, adversarial manipulation, strict latency budgets, and regulatory demands for explainability. Blockchain introduces complementary preventive controls, including tamper‑evident audit trails, shared transaction provenance, decentralized identity support, and smart‑contract automation for policy enforcement. However, blockchain integration creates trade‑offs in throughput, privacy, and governance that necessitate careful partitioning of on‑chain and off‑chain functions. Based on the reviewed evidence, a hybrid security model is proposed that unifies streaming feature engineering, neural risk scoring, an adaptive decision engine, and a permissioned blockchain audit layer to enable verifiable decisions and cross‑institution intelligence sharing.
