Dynamic Cyber Risk Assessment for Industrial Automation: Integrating Digital Governance and Predictive Analytics

Reza Salimi Beni1

Abstract :
This study aimed to develop and empirically evaluate a dynamic cyber risk assessment framework for industrial automation by integrating technical risk indicators, digital governance capability, security control effectiveness, and predictive analytics. A quantitative, applied, longitudinal predictive design was conducted among 278 professionals working in cybersecurity, industrial automation, operational technology, information technology governance, and cyber risk management in Tehran, Iran. Data were collected using a validated multidimensional questionnaire and structured operational cybersecurity indicators obtained from industrial monitoring, vulnerability assessment, incident, and security information systems. The measurement model was assessed using confirmatory factor analysis, Cronbach’s alpha, composite reliability, and average variance extracted. Relationships among study variables were examined using correlation and multivariate regression analyses. Predictive performance was evaluated using logistic regression, random forest, gradient boosting, and extreme gradient boosting models, with out-of-sample assessment based on accuracy, precision, recall, F1 score, sensitivity, specificity, area under the receiver operating characteristic curve, and Brier score. Dynamic cyber risk was positively associated with industrial automation vulnerability (r = 0.69, p < .001) and cyber threat exposure (r = 0.63, p < .001), and negatively associated with digital governance capability (r = -0.57, p < .001), security control effectiveness (r = -0.62, p < .001), and predictive analytics capability (r = -0.49, p < .001). The regression model explained 68.4% of the variance in dynamic cyber risk (R² = 0.684, p < .001). Industrial automation vulnerability was the strongest positive predictor (β = 0.329, p < .001), while security control effectiveness was the strongest negative predictor (β = -0.222, p < .001). Extreme gradient boosting achieved the best predictive performance, with accuracy = 0.889, F1 = 0.874, and AUC = 0.944. The full integrated model increased AUC from 0.873 to 0.944 compared with the technical-only model. Dynamic cyber risk assessment in industrial automation is substantially improved when technical vulnerabilities and threat exposure are evaluated alongside digital governance, security controls, and predictive analytics, demonstrating that integrated machine-learning models provide a more accurate and actionable representation of evolving cyber risk than conventional technical-only approaches.

Keywords : industrial automation cyber risk assessment digital governance predictive analytics cybersecurity operational technology machine learning cyber resilience


زبان مقاله : انگلیسی
محل انتشار :
آدرس وب سایت کنفرانس : https://journaldtai.com
کد IOI مقاله : AEAG-ACAGHK
لینک مستقیم مقاله : https://isnac.ir/AEAG-ACAGHK