QS-DeepGuard: A Next-Generation Cybersecurity Framework for Quantum-Safe Medical Data Protection Using Deep Learning Algorithms and HIPAA-Compliant Architecture

Abstract

The quantum computing revolution is emerging quickly, and is a threat to existing public-key cryptographic systems that protect ePHI. Healthcare organisations are still one of the most targeted sectors in the world with data breaches exposing millions of patient records each year and with the regulations of the Health Insurance Portability and Accountability Act of 1996 (HIPAA). In this paper, we propose a novel, multi-layered cybersecurity framework, called QS-DeepGuard, that combines two post-quantum cryptographic (PQC) algorithms, CRYSTALS-Kyber-1024 and CRYSTALS-Dilithium3, with a diverse ensemble of deep learning models including Long Short-Term Memory (LSTM) networks, Transformer-based intrusion detection modules, and stacked autoencoders for real-time anomaly detection. The framework also includes a federated learning topology that allows for the coordination of multiple healthcare institutions to collectively develop shared threat-detection models without sharing raw patient data, thereby leveraging the benefits of federated learning while addressing privacy concerns. It is evaluated using two publicly available healthcare intrusion datasets (MIMIC-IV and CIC-IDS-2022), the results of which show that QS-DeepGuard can achieve a detection accuracy of 98.76%, an AUC-ROC of 0.9984 and a mean inference latency of 141 ms under full Kyber-1024 encryption, which is a 4.45 percentage-point improvement over the strongest unaided deep learning baseline. A formal HIPAA compliance mapping is included, which shows how each technical component meets the Security Rule safeguards. The cryptographic overhead added by the PQC layer is still within the 0.17ms per transaction, making the framework clinically feasible for real-time electronic health record (EHR) applications.