A Cryptographically Secure and Explainable AI Framework for Automated Health Insurance Claim Processing Using the Insurefusionnet

Abstract

Automated health insurance claim processing has become increasingly important as insurers rely on intelligent systems to handle growing claim volumes and automated decision support systems for claim transparency. However, existing models often suffer from limited interpretability, insufficient data security, and weak generalizability across diverse claim patterns. To address these challenges, this research proposes InsureFusionNet, a hybrid explainable structured ensemble framework integrating heterogeneous models of deep feature learning, uncertainty-aware prediction, interpretable boosting mechanisms, and high-performance gradient-boosting classifiers through a fusion strategy for robust health insurance claim approval classification. Explainable AI techniques, including SHAP and LIME, are incorporated to provide transparent justifications for automated claim approval decisions, thereby enhancing stakeholder trust and accountability. To ensure data privacy and security, the proposed framework integrates AES-256-GCM encryption for confidentiality, elliptic curve cryptography for secure authentication, and SHA-256 hashing for integrity assurance within a permissioned Hyperledger Fabric blockchain, enabling controlled access, tamper-resistant auditability, and trustworthy claim management. Experimental results demonstrate that InsureFusionNet achieves superior performance, attaining an accuracy of 97.24%, precision of 98.41%, recall of 95.89%, and F1-score of 97.11% compared to individual classifiers. Overall, the proposed framework offers a secure, transparent, and reliable solution for secure and explainable automated health insurance claim approval classification and contributes toward responsible and trustworthy AI deployment in healthcare systems.