GEN-SECHEALTH: an AI-powered generative architecture for self-scalable cybersecurity and flexible data privacy protection in intricate healthcare systems

Abstract

The rising levels of cyberattacks on critical infrastructure, especially healthcare systems, require enhanced cybersecurity systems that can offer a high level of detection as well as high data privacy. In this paper, we present GEN-SECHEALTH, an artificial intelligence (AI) based cybersecurity system that uses Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs) to create quality, privacy-aware, synthetic data to enhance anomaly detection in the presence of data scarcity and class imbalance. The artificial data are then used to train an ensemble detection engine using the models of Random Forest, Support Vector Machine (SVM) and Long Short-Term Memory (LSTM) which makes it possible to detect both temporal and static cyber threats. Experimental assessment in various areas is used to show that the proposed framework is effective, reaching 96% accuracy, 94% precision, 93% recall and 0.94 F1-score in healthcare systems, 94% accuracy, 95% recall and 94% precision in financial transaction monitoring, and 91% accuracy and a 0.90 F1-score in smart city IoT settings, with low false positive rates. In each of the assessed scenarios, GEN-SECHEALTH is able to provide end-to-end response times of about 1.5 seconds, thus providing real-time operational demands during peak load. The framework also includes the application of a differential privacy, encryption and access control measures to assist in privacy-compliant deployment in controlled environments. These outcomes support the study that the combination of generative modelling and ensemble learning can increase the detection robustness, scalability, and efficiency without jeopardising data privacy, so GEN-SECHEALTH can be implemented in practise to strengthen the cybersecurity of healthcare and other critical infrastructure.

Ähnliche Arbeiten