NEURAL NETWORK–BASED RISK PREDICTION AND SIMULATION FRAMEWORK FOR MEDICAL IOT CYBERSECURITY: AN ENGINEERING MANAGEMENT MODEL FOR SMART HOSPITALS

Abstract

The rapid digitalization of healthcare through the adoption of Medical Internet of Things (MIoT) technologies has given rise to smart hospital ecosystems that are highly efficient yet increasingly vulnerable to cybersecurity threats. As MIoT devices become integral to patient monitoring, diagnostics, and treatment, the risk of cyberattacks—ranging from ransomware and data breaches to insider threats and Distributed Denial of Service (DDoS) attacks—has grown substantially. In response, this study conducts a structured meta-analysis to evaluate the effectiveness of neural network–based risk prediction and simulation frameworks in securing smart hospital environments. Using the PRISMA 2020 methodology, the review systematically screened and synthesized findings from 112 peer-reviewed studies published between 2010 and 2024, encompassing various experimental setups, real-world hospital case studies, and benchmark datasets. The meta-analysis focused on comparing performance metrics such as detection accuracy, false positive rates, real-time responsiveness, and attack versatility between traditional cybersecurity systems and advanced neural network architectures, including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, and hybrid deep learning models. The findings indicate that neural network–based intrusion detection systems (NN-IDS) consistently outperform rule-based and statistical models, achieving higher accuracy in identifying both known and novel cyber threats. Additionally, these models demonstrate significant reductions in false positive rates and enhanced responsiveness under real-time operational constraints, which are critical for patient safety in clinical environments. These simulation tools support data-driven decision-making and engineering management by forecasting breach impacts, operational disruptions, and compliance risks. Moreover, the adaptability and scalability of NN-IDS across different hospital sizes and digital maturity levels position them as suitable for wide-scale deployment in healthcare systems globally. Overall, this research offers a comprehensive evaluation of neural network–enabled cybersecurity solutions and establishes their practical and strategic value in developing resilient, intelligent, and secure smart hospital infrastructures.