A process model for dynamic analysis and prediction of information security risks for personnel

Abstract

The article addresses the problem of dynamic assessment and forecasting of information security risks driven by the growing role of the human factor amid business-process digitalization, hybrid work models, and a changing access context. The purpose of the study is to improve the accuracy and timeliness of risk management for personnel, enhance access controllability through adaptive policies, and advance audit transparency by integrating an RBAC-blockchain into a closed self-learning loop. The object of the study is the process of dynamic information-security risk analysis in corporate systems with personalized consideration of user behavior. The subject of the study comprises the methods and procedures for constructing a multidimensional feature vector, building a user digital twin, designing risk-adaptive access policies, and implementing system self-learning mechanisms. The authors emphasize that static access-control approaches and periodic audits do not match the dynamics of threats and the contextual nature of resource use. The article analyzes the contemporary components of the process model: (f₁) construction of a multidimensional resource-classification matrix; (f₂–f₃) collection, unification, and normalization of behavioral and technical data into the feature vector Q; (f₄) forecasting risky events using a user digital twin with access transactions recorded on an RBAC-blockchain; (f₅–f₆) generation of adaptive countermeasures and delivery of personalized policies and training content; and (f₇–f₈) Fback feedback collection and self-learning with adjustment of weights, models, and access rules. It is shown that combining statistical methods, machine-learning algorithms, and immutable blockchain logging ensures reproducible auditing, reduces the “risk window,” and supports continuous trust validation in line with Zero Trust principles. A scheme for triggering countermeasures is proposed based on the probability matrix R and the resource’s criticality class. Procedures for fine-tuning and transfer learning are described to keep models current without excessive computational costs. Particular attention is paid to personalized dashboards and multichannel delivery of recommendations that shorten user response time. The importance of qualitative Fback metrics (e.g., user satisfaction and content clarity) is emphasized for revealing elements of security culture. Thus, applying the developed process model establishes an “analysis–forecast–action–feedback–self-correction” cycle that improves the accuracy of risk assessment, enhances response timeliness, and advances transparency in access governance. The results can be integrated into SIEM/UEBA environments, access-management systems, and corporate programs for improving personnel cyber literacy.