Advanced modeling and management strategies for nuclear and radiological incidents: from decision support to adaptive governance

1 Introduction

Nuclear and radiological incidents represent low-probability but high-consequence events with potentially severe impacts on public health, the environment, critical infrastructure, and societal trust. Over the past decades, substantial efforts at national and international levels have focused on strengthening Emergency Preparedness and Response (EPR) through improved regulatory frameworks, monitoring networks, modelling capabilities, and coordinated response arrangements. These efforts have significantly enhanced the technical capacity to detect, characterize, and assess radiological hazards (Renn and Levine, 1991; Covello, 2003; World Health Organization, 2017).

Atmospheric dispersion models, dose assessment tools, and consequence analysis frameworks have become essential components of contemporary EPR, providing analytical insight into the potential evolution and impacts of radiological release. In parallel, digital platforms and Decision Support Systems (DSS) have been developed to integrate modelling outputs with monitoring data, operational procedures, and communication tools. Collectively, these developments have transformed the analytical landscape of nuclear and radiological emergency management (Hanna et al., 1982; Ehrhardt et al., 2000; IAEA, 2018).

Despite these advances, experience from real-world events and large-scale international exercises have demonstrated that accurate modelling alone does not guarantee effective emergency management. Decision-makers are often confronted with rapidly evolving situations characterized by uncertainty, incomplete information, conflicting objectives, and intense organizational and societal pressures where analytical outputs must be interpreted, contextualized, and reconciled with competing priorities.

As a result, the effectiveness of modelling and decision-support depends not only on technical performance, but also on how these capabilities are embedded within organizational structures, governance arrangements, and human-centered decision processes.

The literature on EPR reflects a growing recognition of this gap between analytical capability and operational decision-making. While significant attention has been devoted to improving model fidelity, computational speed, and data integration, comparatively less emphasis has been placed on how modelling outputs are used in practice, how uncertainty is communicated, and how decision-support systems interact with institutional authority and human judgement. This has led to recurring tensions between preparedness-oriented analytical tools and the realities of response-phase decision-making (IAEA, 2021; OECD Nuclear Energy Agency (NEA), 2014; Comfort et al., 2010).

In response to these challenges, recent research increasingly frames EPR as a socio-technical and organizational problem rather than a purely technical one. Concepts such as complex adaptive systems, human-in-the-loop decision-making, and hybrid operational frameworks have been introduced to better capture the dynamic interactions between analytical tools, organizational processes, and human actors. Within this context, Hybrid Emergency Operations Centers (Hybrid EOCs) have emerged as a conceptual response to the need for integrative environments that connect modelling, decision-support, coordination, and communication across institutional boundaries.

The objective of this article is to critically and integratively review advanced modelling and management strategies for nuclear and radiological incidents, with particular attention to how analytical capabilities are embedded within decision-making and governance frameworks. Rather than focusing solely on modelling techniques, the paper adopts an integrative perspective that encompasses DSS, organizational processes, and human-in-the-loop decision-making (French et al., 2009; Endsley, 1995; Parasuraman and Riley, 1997).

Specifically, the review:

• Analyses nuclear and radiological EPR through the lens of complex adaptive systems (Holling, 2001; Manson, 2001; Perrow, 1999; Comfort, 1999);
• Reviews the evolution of modelling and decision-support approaches and their limitations in operational contexts;
• Discusses hybrid operational concepts, including Hybrid Emergency Operations Centers (Hybrid EOCs), as integrative governance layers (FEMA, 2017; ISO, 2018; Turoff et al., 2004);
• Identifies key gaps and emerging research directions aimed at strengthening the translation of advanced modelling capabilities into effective management strategies (Bankes, 1993; Deb, 2001; Walker et al., 2013; Lempert, 2019).

By synthesizing insights from modelling, decision-support research, and emergency management practice, this review aims to clarify where the principal challenges of contemporary EPR now reside and to articulate conceptual directions for future research and practice. In doing so, it positions advanced modelling not as an end in itself, but as an enabling component within integrated, adaptive emergency management ecosystems.

2 Methodological approach of the review

This article is structured as a critical and integrative review of advanced modelling and decision-support approaches in nuclear and radiological emergency preparedness and response (EPR). The purpose of the review is not to provide an exhaustive inventory of existing models or systems, nor to conduct a systematic or quantitative meta-analysis, but to synthesize and interpret the literature to identify recurring patterns, structural limitations, and emerging challenges in the operational use of advanced analytical capabilities within complex emergency environments.

This methodological choice reflects the nature of the research problem. Nuclear and radiological emergencies are characterized by the interaction of technical, organizational, and human factors, which cannot be meaningfully examined through a single disciplinary lens or reduced to purely technical performance metrics.

A critical and integrative review therefore provides an appropriate framework for examining how modelling and decision-support systems function in practice, how they are embedded in governance structures, and how they interact with human decision-making under conditions of uncertainty and stress.

While not intended as a systematic review in the PRISMA sense, this work follows a transparent and structured analytical logic suited to its integrative and conceptual objectives.

2.1 Review scope and literature sources

The literature reviewed for this study was drawn from a combination of peer-reviewed scientific journals, conference proceedings, international technical reports, and policy-relevant guidance documents, complemented by the author’s professional experience in nuclear and radiological emergency preparedness and response, used to contextualize and interpret the reviewed material rather than as a primary source of evidence. Primary sources included publications indexed in major scientific databases, such as Scopus and Web of Science, complemented by authoritative international guidance and standards relevant to EPR.

The literature primarily covers publications from the early 2000s onward, reflecting the maturation of computational modelling and DSS in EPR.

The scope of the review spans multiple, interrelated domains, including:

• Atmospheric dispersion modelling, dose assessment, and consequence analysis;
• Decision-support systems (DSS) used in nuclear and radiological emergencies;
• Emergency management, organizational coordination, and governance frameworks;
• Human factors, uncertainty, stress, and decision-making in high-consequence environments;
• Large-scale exercises and preparedness initiatives as mechanisms for validation and learning.

Sources focusing exclusively on model formulation or computational performance were considered only where they contributed to understanding the operational role of modelling in preparedness, response, or decision-making contexts.

2.2 Selection logic and analytical orientation

Rather than applying rigid, predefined analytical categories or formal inclusion–exclusion criteria typical of systematic reviews, the review follows a thematically driven and comparative analytical logic. The selection of sources was guided by their relevance to the operational use of modelling and decision-support in EPR, with particular attention to how analytical outputs are interpreted, communicated, and integrated into decision-making processes.

The literature was examined comparatively to identify:

• How modelling and DSS are conceptualized across preparedness and response phases;
• How uncertainty is represented, communicated, and acted upon;
• How analytical outputs align or fail to align with decision timelines, organizational mandates, and coordination requirements;
• How exercises and training activities shape the practical use of analytical tools.

This approach enables the identification of cross-cutting themes and tensions that persist across different technical systems, institutional settings, and national contexts.

2.3 Conceptual lens: complex adaptive systems

The analytical synthesis is guided by the conceptual lens of Complex Adaptive Systems (CAS). For example, feedback between evolving hazard assessments, organizational constraints, and public communication decisions illustrates how adaptive behavior emerges during emergency responses.

This perspective is not employed as a formal mathematical framework, but as an interpretative tool for understanding emergency preparedness and response as a dynamic socio-technical system. Within this lens, nuclear and radiological emergencies are viewed as emergent phenomena arising from interactions among technical tools, organizational structures, human judgement, and external stressors (perturbation factors).

Applying a CAS perspective allows the review to move beyond linear or deterministic interpretations of emergency management and to examine how feedback loops, adaptation, non-linear effects, and institutional interactions shape the effectiveness of modelling and decision-support systems in practice.

2.4 Nature and output of the review

The outcome of this methodological approach is a conceptual integration, rather than a prescriptive model or technical solution. Insights derived from the reviewed literature are used to:

• Characterize structural limitations of prevailing scenario-based and threshold-driven decision-support paradigms;
• Identify governance and human-in-the-loop challenges that constrain the operational use of advanced modelling;
• Derive conceptual implications for adaptive, governance-aware decision-support frameworks.

Within this context, concepts such as Hybrid Emergency Operations Centers (Hybrid EOCs) are introduced as conceptual outputs of the review.

This concept serves to bridge analytical capabilities and institutional decision-making, rather than to replace existing organizational authority or human judgement. This methodological positioning establishes the foundation for the subsequent analysis in Section 3, which examines how advances in modelling, decision-support systems, and organizational frameworks have shaped contemporary EPR practice and how these elements interact within complex emergency environments.

3 EPR as a complex adaptive system

EPR exhibits the defining characteristics of a CAS. Such systems are composed of multiple interacting subsystems, operate across different spatial and temporal scales, and display non-linear behavior, feedback loops, and emergent outcomes that cannot be fully predicted from the behavior of individual components alone.

In the context of nuclear and radiological incidents, system behavior emerges from the interaction between technical processes (e.g., source term evolution, atmospheric dispersion, dose accumulation), organizational structures (e.g., regulatory authorities, emergency services, operators), human decision-makers, and societal dynamics, including public perception and trust. These interactions are further shaped by external perturbations such as evolving meteorological conditions, infrastructure disruptions, cross-border information flows, and political or media pressures (Renn and Levine, 1991; Covello, 2003; World Health Organization, 2017).

3.1 Interacting subsystems and system boundaries

From a systems perspective, EPR can be understood as a constellation of interdependent subsystems rather than a single, centrally controlled entity. Core subsystems typically include:

• Technical subsystems, encompassing monitoring networks, modelling tools, and analytical platforms
• Organizational and institutional subsystems, including regulatory authorities, emergency management bodies, and response organizations
• Human and cognitive subsystems, involving decision-makers, responders, and expert advisors
• Societal subsystems, encompassing affected populations, communication channels, and trust relationships.

The boundaries between these subsystems are inherently porous. Information, decisions, and actions continuously cross institutional and functional interfaces, often under severe time constraints. As a result, local decisions can have system-wide effects, and small perturbations—such as delays in information sharing or misalignment of assumptions may propagate non-linearly through the system.

Figure 1 provides a conceptual representation of nuclear and radiological EPR as a complex adaptive system, illustrating the interacting subsystems and feedback mechanisms discussed in this section (Holling, 2001; Manson, 2001; Perrow, 1999; Comfort, 1999).

3.2 Non-linearity, feedback and emergence in EPR

A defining feature of complex adaptive systems is non-linearity whereby proportional inputs do not necessarily lead to proportional outputs. In nuclear and radiological emergencies, non-linear effects arise from multiple sources, including threshold-based protective actions, cascading organizational responses, and public behavioral reactions to official communications.

Feedback loops further complicate system behavior. For example, model outputs influence protective action decisions, which in turn affect monitoring strategies, public behavior, and subsequent data inputs. Similarly, communication strategies may alter public perception and compliance, thereby influencing exposure pathways and operational priorities. These feedback mechanisms can amplify or dampen system responses in ways that are difficult to anticipate in advance.

Emergent phenomena such as coordination breakdowns, decision paralysis, or conversely, rapid collective adaptation often arise from these interactions. Importantly, such outcomes are not necessarily the result of failures at the component level, but rather of misalignments between subsystems operating under different constraints, objectives, and temporal rhythms (Bankes, 1993; Deb, 2001; Walker et al., 2013; Lempert, 2019).

3.3 Implications for modelling and decision-support

Conceptualizing EPR as a complex adaptive system has important implications for the role of modelling and decision-support tools. While physical and computational models are essential for understanding hazard evolution and potential impacts, they represent only a subset of the information required for effective emergency management. Model output must be interpreted, contextualized, and reconciled with organizational capabilities, legal frameworks, and societal considerations.

Deterministic and scenario-based decision-support approaches, which assume relatively stable system behavior and predefined response pathways, may be insufficient to capture the adaptive dynamics of real-world emergencies. In complex systems, uncertainty is not merely a lack of information to be reduced, but an intrinsic property of system behavior that must be managed (IAEA, 2021; OECD Nuclear Energy Agency (NEA), 2014; Comfort et al., 2010).

Here, the term “deterministic” does not refer to the physical or mathematical properties of dispersion or dose models, which remain inherently uncertain and often probabilistic, particularly in the early phase of an event. Instead, it refers to the structure of decision-support logic, whereby uncertain analytical outputs are mapped onto predefined thresholds, workflows, and procedural decision pathways within institutional response frameworks.

This perspective suggests that decision-support tools should be designed not only to provide accurate predictions, but also to support adaptive reasoning, flexibility, and learning under uncertainty. Rather than seeking to eliminate complexity, effective EPR frameworks must enable institutions to operate coherently within it.

3.4 From system characterization to governance

Recognizing EPR as a complex adaptive system shifts the focus from optimizing individual components to governing interactions and adaptation. Governance, in this sense, refers to the mechanisms by which institutions coordinate actions, align interpretations, and maintain accountability in the face of uncertainty and change. This governance-oriented interpretation of emergency management as a systemic and adaptive process is consistent with more recent disaster risk literature, which emphasizes systemic thinking and the management of interactions across institutional and sectoral boundaries rather than optimization of isolated system components (Alexander and Pescaroli, 2020).

This shift provides the conceptual foundation for hybrid operational approaches that integrate modelling, decision-support, organizational workflows, and communication processes within a coherent framework. As discussed in the following section, Hybrid Emergency Operations Centers (Hybrid EOCs) represent one such approach, offering a structured environment for managing complexity while preserving human authority and institutional responsibility (FEMA, 2017; ISO, 2018; Turoff et al., 2004).

4 Modelling and decision-support in EPR

Modelling and decision-support have long been central pillars of EPR. Over time, substantial investments have been made in the development of physical models, computational tools, and integrated platforms designed to support emergency decision-making. While these capabilities have significantly improved situational awareness and analytical depth, their role within complex emergency management processes requires careful examination.

This section examines modelling and decision-support in nuclear and radiological emergency preparedness and response through a dual analytical lens.

First, it reviews the prevailing theoretical and technical paradigms reflected in the literature, including advances in dispersion modelling, consequence assessment, and decision-support system design.

Second, building on this analysis, the section develops an applied and conceptual synthesis that interprets how these capabilities are operationalized in practice and identifies structural limitations that motivate the need for integrative management and governance approaches.

Rather than treating modelling and decision-support as standalone technical functions, the analysis emphasizes their role within broader emergency management systems characterized by organizational complexity, uncertainty, and multi-actor coordination.

This dual perspective allows the review to distinguish between established analytical paradigms and the conceptual outputs derived from the literature, which inform the discussion of hybrid operational frameworks and adaptive decision-making later in the section.

4.1 Evolution of dispersion modelling in EPR

From a theoretical perspective, dispersion and consequence models have traditionally been developed to support preparedness-oriented planning and scenario analysis.

Historically, the primary purpose of dispersion model outputs has been to support preparedness activities, rather than to provide prescriptive guidance for real-time operational decision-making. Early modelling efforts were designed to inform planning, zoning, consequence assessment, and the definition of protective action strategies under a wide range of hypothetical scenarios. These applications were inherently exploratory and offline in nature, allowing decision-makers to examine potential outcomes without the pressures of time-critical response (Turner, 1970; Hanna et al., 1982; Ehrhardt et al., 2000; IAEA, 2018).

The earliest generation of dispersion models prioritized analytical simplicity and interpretability. Steady-state Gaussian plume models, for example, enabled rapid estimation of contaminant dispersion under idealized assumptions. While these models involved substantial simplifications, their transparency and computational efficiency made them well suited for preparedness tasks such as screening assessments and conservative planning analyses. Their limitations were generally well understood, and their outputs were interpreted as indicative rather than definitive representations of real-world conditions.

Subsequent developments introduced time-dependent formulations and closer coupling with meteorological data. Puff-based models and dynamic atmospheric transport simulations improved the representation of temporal variability and changing environmental conditions. These advances enhanced the realism of dispersion assessments and expanded the range of scenarios that could be explored during preparedness planning and exercises. Importantly, however, these models continued to serve primarily as analytical tools for scenario exploration, rather than as mechanisms for operational decision automation. Even as uncertainty representations became more explicit, model outputs remained inputs to expert judgement rather than substitutes for it (IAEA, 2021; OECD Nuclear Energy Agency (NEA), 2014; Comfort et al., 2010).

The integration of dispersion models into comprehensive consequence assessment platforms marked a further step in this evolution. Systems that combined source term estimation, atmospheric dispersion, dose calculation, and evaluation of potential countermeasures offered an end-to-end analytical perspective on emergency consequences. Such platforms significantly strengthened preparedness by enabling consistent scenario analysis, comparison of protective strategies, and support for training and exercises. At the same time, their increasing sophistication introduced new interpretive challenges, including multiple outputs, sensitivity to assumptions, and the potential for divergent analytical results across institutions.

As modelling systems became more integrated and computationally efficient, their output became increasingly available in near-real-time environments. This availability has, at times, encouraged expectations that dispersion models should directly inform or even guide operational decisions during response. This shift has not altered the original conceptual purpose of these models, but it has exposed a growing tension between analytical capability and decision-making context. Outputs developed for preparedness and scenario exploration may be difficult to interpret under the uncertainty, time pressure, and organizational complexity that characterize real emergencies.

From a complex adaptive systems perspective, this tension is not a failure of modelling, but a reflection of the evolving role assigned to analytical tools within emergency management. Decision-making during response emerges from interactions among technical assessments, organizational structures, human judgement, and societal considerations. Dispersion models provide essential situational insight, but their value during response depends on how their outputs are interpreted, communicated, and integrated into coordinated action rather than on analytical precision alone.

This evolution underscores a critical shift in emphasis: advances in dispersion modelling have primarily strengthened preparedness capabilities, while their effective use during response requires governance frameworks that support interpretation, coordination, and human-centered decision-making. Hybrid operational concepts, such as Hybrid EOC, address this challenge by embedding modelling outputs within structured environments that facilitate dialogue between analysts and decision-makers, preserve institutional accountability, and accommodate uncertainty as an inherent feature of complex emergencies.

This distinction reflects a governance and safety philosophy embedded in international emergency preparedness and response guidance, where urgent response decisions are based on predefined protective strategies, conservative criteria, and expert judgement, with modelling serving an exploratory and contextual role rather than a prescriptive decision-making function.

4.2 Decision-support systems and their limitations

In the literature, Decision-Support Systems (DSS) are commonly framed as integrative tools designed to operationalize modelling outputs within predefined response procedures. Moreover, DSS has been developed to bridge the gap between analytical modelling and operational decision-making. Building on dispersion modelling, monitoring data, and predefined response arrangements, these systems typically integrate analytical outputs, procedural guidance, and visualization tools within a common operational interface. Their primary contribution lies in supporting situational awareness, facilitating information sharing, and promoting a common operational picture among emergency actors (Turner, 1970; Hanna et al., 1982; Ehrhardt et al., 2000; IAEA, 2018).

Historically, the design of many DSS reflects their close connection to preparedness activities. Scenario-based planning, emergency zoning, and predefined protective action strategies have shaped DSS logic, resulting in systems that are largely procedure-oriented and aligned with regulatory thresholds. In this context, DSS support consistency, traceability, and compliance with established emergency plans qualities that are essential for coordinated response across institutions.

However, this scenario-driven and procedural orientation also introduces limitations when DSS are used in dynamic response contexts. Real-world nuclear and radiological emergencies rarely unfold according to predefined scripts. Instead, decision-makers must contend with evolving information, competing objectives, resource constraints, and interactions among multiple actors operating under different mandates. In such conditions, decision-making extends beyond selecting actions from predefined pathways and involves continuous reasoning about timing, feasibility, trade-offs, and coordination.

From a complex adaptive systems perspective, these limitations become particularly salient. DSS are often optimized to support linear decision pathways, whereas response processes emerge from non-linear interactions among technical assessments, organizational structures, and human judgement. As a result, DSS may struggle to accommodate deviations from anticipated scenarios or to support adaptive reasoning when multiple response options remain plausible.

A further challenge concerns the representation and interpretation of uncertainty. Many DSS present modelling outputs and projections in ways that may implicitly convey a degree of precision or confidence that is not fully justified, particularly during the early phases of an incident. When uncertainty is not explicitly communicated or explored, users may develop an undue reliance on specific outputs, increasing the risk of cognitive fixation or automation bias. This risk is amplified under time pressure and stress, when decision-makers may seek apparent certainty in analytical results.

These observations do not diminish the value of decision-support systems. On the contrary, they underscore the need to reconsider their role within emergency management frameworks. Rather than functioning primarily as instruments for executing predefined response logic, DSS should be understood as tools that support adaptive reasoning, enable exploration of alternative assumptions and outcomes, and inform human judgement under uncertainty. Their effectiveness depends not only on analytical sophistication, but also on how they are embedded within organizational processes and governance structures that preserve accountability and support coordinated action (French et al., 2009; Endsley, 1995; Parasuraman and Riley, 1997).

4.3 Hybrid Emergency Operations Centers as an integrative layer

Building on the limitations identified in prevailing modelling and decision-support paradigms, this subsection introduces the Hybrid Emergency Operations Center (Hybrid EOC) as a conceptual and applied synthesis emerging from the reviewed literature.

The Hybrid EOC is not proposed here as a standardized operational model or technological solution, but as a synthesized conceptual construct derived from documented practices observed in large-scale exercises, digitally supported emergency operations centers, and hybrid coordination arrangements that combine physical command structures with remote expert participation.

In response to the growing complexity of nuclear and radiological emergency management, hybrid operational concepts have emerged that seek to better integrate analytical modelling, decision-support systems, organizational workflows, and communication processes. Among these, the concept of the Hybrid EOC has gained increasing attention as a means of addressing the gap between preparedness-oriented analytical capabilities and the realities of operational response (FEMA, 2017; ISO, 2018; Turoff et al., 2004).

Hybrid EOCs combine physical command structures with digital environments that enable real-time access to model outputs, monitoring data, situational information, and collaborative tools. This hybridization allows geographically distributed actors including analysts, decision-makers, and supporting experts to participate in coordinated response processes while preserving established institutional roles, mandates, and chains of authority. Importantly, Hybrid EOCs are not defined by specific technologies or software platforms, but by their function as integrative environments that support coordination and interpretation in complex emergency contexts.

The emergence of Hybrid EOCs can be understood as a response to the limitations identified in preparedness-oriented modelling and scenario-driven decision-support systems. While dispersion models and DSS provide essential analytical insight, their outputs require contextual interpretation, particularly during response phases characterized by uncertainty, time pressure, and competing objectives. Hybrid EOCs address this challenge by embedding analytical tools within structured operational settings that facilitate dialogue between technical experts and decision-makers, rather than treating modelling outputs as prescriptive inputs to action.

From a governance perspective, the Hybrid EOC functions as a boundary-spanning layer between technical analysis and institutional decision-making. It provides a shared space in which modelling results, expert judgement, operational constraints, and communication considerations can be jointly examined. By making assumptions, uncertainties, and decision rationales explicit, Hybrid EOCs enhance transparency, traceability, and accountability—attributes that are essential for legitimate decision-making in high-stakes emergency situations.

Within the framework of complex adaptive systems, the role of the Hybrid EOC is not to impose centralized control or to eliminate uncertainty. Instead, it supports coordinated adaptation by aligning information flows, decision processes, and organizational interactions across multiple actors. This alignment enables institutions to respond coherently to evolving situations without constraining the flexibility required to address unforeseen developments or compound emergencies.

The relevance of Hybrid EOCs becomes particularly pronounced in multi-actor and cross-border contexts, where differences in organizational mandates, analytical assumptions, and communication practices can hinder coordination. By providing common reference points, shared situational awareness, and mechanisms for continuous interaction, Hybrid EOCs help build trust and facilitate convergence of interpretations, even when full agreement on analytical details is not achievable. In this sense, Hybrid EOCs represent a pragmatic response to the governance challenges posed by EPR as complex adaptive systems (Holling, 2001; Manson, 2001; Perrow, 1999; Comfort, 1999).

Figure 2 presents the conceptual architecture of a Hybrid EOC, highlighting its role as an integrative environment connecting physical command structures, digital decision-support, and human-centered coordination.

4.4 Applicability of existing decision-support systems to compound and cascading emergencies

EPR frameworks have traditionally been developed around single-hazard assumptions, in which the radiological event itself represents the primary driver of system behavior. This orientation is reflected in the design of the existing decision-support systems, whose architectures, workflows, and data models are optimized for radiological source terms, dispersion processes, and dose-based decision criteria. Such systems have proven highly effective for preparedness activities and for structured response within anticipated radiological scenarios.

However, recent experience, scenario analyses, and large-scale exercises increasingly underscore the relevance of compound and cascading emergencies, in which nuclear or radiological incidents occur alongside other disruptive events such as earthquakes, floods, extreme weather, or large-scale infrastructure failures. In these situations, the radiological hazard interacts with broader system stressors, generating non-linear effects and amplifying overall complexity. As a result, the assumptions underpinning single-hazard response frameworks are frequently challenged.

From a technical perspective, many existing DSS possess a degree of extensibility that allows the integration of additional data layers or external inputs. Geographic information systems, for example, can incorporate information on infrastructure damage, population displacement, accessibility constraints, or weather impacts. Nevertheless, these extensions typically operate at the level of visualization or contextual awareness. The core analytical logic of most DSS remains centered on radiological consequence assessment, with limited capability to explicitly represent interdependencies among multiple hazards, critical infrastructures, resource availability, and organizational functions (French et al., 2009; Endsley, 1995; Parasuraman and Riley, 1997).

The limitations of single-hazard DSS become more pronounced at the decision-making level. Compound emergencies introduce competing and, at times, conflicting priorities, such as balancing radiological protection with life-saving rescue operations, infrastructure stabilization, or large-scale evacuation under degraded conditions. These trade-offs cannot be resolved through radiological metrics or predefined thresholds alone. Instead, they require contextual judgement, cross-sector coordination, and adaptive sequencing of actions as conditions evolve and uncertainty persists.

Organizational factors further complicate the use of DSS in compound emergency contexts. Many systems implicitly assume the availability of monitoring networks, reliable communication channels, and stable command structures. In compound events, these assumptions may no longer hold. Earthquake-induced damage, for instance, can disrupt sensor networks, impair data transmission, or physically constrain response teams, thereby undermining the reliability and timeliness of model-driven outputs. In such circumstances, DSS output must be interpreted cautiously and situated within a broader understanding of operational constraints.

Within the framework of complex adaptive systems, compound and cascading emergencies highlight the inherent limits of single-domain decision-support tools. While existing DSS remain indispensable for providing domain-specific analytical insight, their operational relevance in complex emergencies depends on their integration within broader coordination and governance mechanisms capable of accommodating multiple interacting hazards. Hybrid operational environments, such as Hybrid EOC, provide such mechanisms by enabling the joint interpretation of analytical outputs, organizational constraints, and evolving priorities across sectors and institutions (Holling, 2001; Manson, 2001; Perrow, 1999; Comfort, 1999).

In this context, the challenge is not to replace existing DSS, but to reposition them as components of an integrated decision environment. By embedding radiological decision-support within governance frameworks that support coordination, transparency, and human-centered judgement, emergency response systems can better navigate the complexity of compound and cascading emergencies without over-reliance on single-hazard analytical logic.

This observation reinforces the importance of integrative operational environments, such as Hybrid Emergency Operations Centers, which can contextualize radiological decision-support within a wider emergency management landscape. Rather than replacing existing DSS, such environments enable their outputs to be interpreted alongside information from other hazard domains, organizational constraints, and evolving situational demands.

Within the framework of complex adaptive systems, compound and cascading emergencies expose the inherent limits of single-domain decision-support tools. While existing DSS remain indispensable for providing radiological and domain-specific analytical insight, their operational relevance in complex emergencies depends on their integration within broader coordination and governance mechanisms capable of accommodating multiple interacting hazards, organizational constraints, and evolving priorities. Hybrid operational environments, such as Hybrid Emergency Operations Centers, provide such mechanisms by enabling the joint interpretation of analytical outputs, contextual information, and decision constraints across sectors and institutions. At the same time, these challenges highlight the need for empirical environments in which the interaction between modelling, decision-support, organizational processes, and human judgement can be systematically examined an issue that is addressed through scenario-based planning and large-scale exercises. The effectiveness of warning and communication systems during radiological emergencies further depends on the resilience of national infrastructures, as illustrated by studies of population warning systems designed for radiation accidents (Kavan and Stýblová, 2025).

Figure 3 illustrates the conceptual gap between single-hazard decision-support systems and the operational realities of compound and cascading emergencies. While radiological modelling and DSS remain essential analytical tools, their effectiveness in complex emergency contexts is conditioned by organizational coordination, governance structures, and the ability to integrate multiple hazard dimensions.

These limitations are therefore not primarily technical in nature, but governance-related, reflecting the challenges of coordinating analytical insight, organizational mandates, and competing priorities within complex adaptive emergency environments.

4.5 Scenario-based planning and large-scale exercises as validation mechanisms

Many of the coordination practices associated with Hybrid EOC are already implicitly exercised in large-scale international exercises, where physical command structures are complemented by remote expert participation and digitally mediated decision-support environments.

Scenario-based planning and large-scale exercises constitute essential mechanisms for validating advanced modelling approaches and integrated emergency management frameworks. While analytical tools and decision-support systems can be extensively tested in laboratory or desktop environments, their operational relevance becomes evident only when they are exposed to realistic conditions of uncertainty, time pressure, and multi-actor coordination. Exercises therefore function not merely as preparedness activities, but as empirical environments in which the interaction between modelling, decision-support, and management can be systematically examined.

Large-scale exercises are particularly valuable because they simulate dynamic and evolving scenarios rather than static, predefined sequences. Such scenarios require participants to reassess assumptions, update analytical inputs, and adapt decisions as new information becomes available. This dynamic character closely mirrors operational reality, in which early information is often incomplete, and modelling outputs must be interpreted iteratively rather than treated as definitive (IAEA, 2021; OECD Nuclear Energy Agency (NEA), 2014; Comfort et al., 2010).

From a conceptual perspective, exercises provide a critical test of integration. They reveal whether modelling outputs are produced within operationally relevant timeframes, whether decision-support logic aligns with management needs, and whether information flows support coherent situational awareness across institutions. Experience consistently shows that limitations frequently arise not from model accuracy, but from misalignment between analytical outputs and decision timelines, organizational responsibilities, or communication practices.

The increasing use of digital exercise platforms has enhanced the capacity of exercises to serve as validation mechanisms. Such platforms enable the structured delivery of scenario injects, real-time information updates, and coordinated participant interactions within shared digital environments. By synchronizing modelling outputs, situational data, and decision points, these platforms support more realistic testing of integrated operational concepts, including hybrid configurations that combine physical coordination with remote participation.

Experience from large-scale international exercises, including recent ConvEx-3 activities, has underscored the importance of shared reference frameworks. Exercises reveal how divergent modelling assumptions, thresholds, or visualization practices can complicate coordination, particularly in multinational contexts. Conversely, scenarios based on harmonized assumptions and common visualization approaches facilitate faster convergence of interpretations and decisions, highlighting the role of exercises in identifying interoperability challenges before real events occur.

Exercises also illuminate the operationalization of uncertainty. Rather than treating uncertainty solely as a technical attribute of models, scenario-based exercises expose how uncertainty is perceived, communicated, and acted upon by decision-makers. Exercises that explicitly incorporate uncertainty into scenario design enable organizations to practice adaptive decision-making and to develop tolerance for ambiguity while maintaining accountability capabilities that are difficult to cultivate through analytical testing alone.

Beyond validation, large-scale exercises function as mechanisms for organizational learning. Structured evaluation processes allow participants to reflect on how modelling outputs and decision-support tools were used, misunderstood, or disregarded. When systematically captured and analyzed, these insights inform improvements in model design, decision logic, interfaces, and management procedures, contributing to the co-evolution of technical and organizational capabilities.

At the same time, exercises have inherent limitations as validation tools. No exercise can fully replicate the political, social, and emotional pressures of real emergencies, and repeated exposure to similar scenarios may encourage scripted responses rather than genuine adaptation. Recognizing these limitations is essential to avoid overinterpreting exercise performance as proof of operational readiness.

Scenario-based planning and large-scale exercises play a central role in validating advanced modelling and integrated management frameworks. They expose integration gaps that are not apparent in technical testing, support the refinement of hybrid operational concepts, and foster continuous learning, thereby forming a critical bridge between conceptual development and operational practice in nuclear and radiological EPR.

Beyond their role as validation mechanisms, large-scale exercises also expose the cognitive and organizational pressures under which decisions are made. By placing decision-makers in dynamic, uncertain, and time-constrained environments, exercises provide valuable insight into how modelling outputs and decision-support tools are interpreted and acted upon under stress.

As illustrated in Figure 4, scenario-based exercises function as critical validation environments for advanced modelling and decision-support systems within complex adaptive emergency management systems. Exercises do not merely test analytical accuracy, but reveal how modelling outputs are interpreted, communicated, and acted upon under realistic operational conditions. By combining dynamic scenarios, evolving information, and distributed participation, exercises enable the assessment of integration gaps that are not apparent through technical testing alone, particularly at the interface between analytical tools, governance structures, and human judgement.

4.6 Decision-making under uncertainty and stress

Decision-making in nuclear and radiological emergencies is inherently shaped by uncertainty, time pressure, and elevated cognitive demands. Unlike routine operational contexts, emergency decision-makers must act on incomplete, evolving, and sometimes contradictory information while coordinating across technical, organizational, and political domains. In such conditions, the effectiveness of advanced modelling and decision-support systems depends not only on their analytical capabilities, but also on how they interact with human judgement and organizational processes under stress.

Uncertainty arises from multiple sources, including incomplete knowledge of the source term, variability in meteorological conditions, limitations of monitoring data, and structural assumptions embedded within models. Although contemporary modelling approaches increasingly quantify uncertainty through probabilistic outputs and ensemble techniques, the principal challenge lies in integrating uncertainty into decision-making processes. Decision-makers often require clear guidance precisely when uncertainty is greatest, particularly during the early phase of an incident (Klein, 1998; Kahneman, 2011; Flin et al., 2008).

Stress amplifies these challenges by affecting attention, information processing, and risk perception. High stakes, compressed timelines, and external pressures such as media scrutiny and political expectations can encourage reliance on heuristics, prior experience, or institutional norms rather than on detailed analytical assessment. While such adaptive behaviors may support rapid action, they also increase susceptibility to cognitive biases, including anchoring, confirmation bias, and premature closure of decision options.

From a technical perspective, the representation and communication of uncertainty play a decisive role in shaping decision outcomes. Advanced modelling systems can generate complex probabilistic information, but without appropriate structuring and visualization, such information may overwhelm users or be disregarded altogether. Decision-support systems must therefore mediate between analytical completeness and cognitive usability, prioritizing representations that support comparison, trend recognition, and exploration of alternatives rather than exhaustive detail.

The interaction between human judgement and computational support remains a central tension. Expert judgement provides contextual understanding, ethical consideration, and adaptive reasoning that models cannot replicate. At the same time, computational tools offer consistency, speed, and the capacity to explore a wide range of scenarios. Effective decision-making under uncertainty requires a balanced interaction in which models inform but do not dictate decisions, and human judgement remains accountable and transparent, as illustrated by the human-in-the-loop paradigm (Klein, 1998; Kahneman, 2011; Flin et al., 2008).

Recent empirical and practice-oriented studies further highlight the ethical dimensions of radiological emergency response, particularly at the operational level of first responders. For example, analyze preparedness and ethical challenges faced by fire and rescue services during radiological response operations, emphasizing the tension between procedural compliance, situational uncertainty, and individual responsibility under time pressure (Kavan and Stýblová, 2025).

Organizational context significantly influences how uncertainty and stress are managed. Clear decision authority, defined roles, and trusted communication channels can mitigate cognitive load and support timely action. Conversely, ambiguous mandates, fragmented information flows, or competing institutional priorities can exacerbate stress and delay decisions. Organizations that routinely engage with modelling and decision-support tools through training and exercises tend to demonstrate greater confidence and competence in their use during crises.

Hybrid operational frameworks provide structural mechanisms to support decision-making under uncertainty and stress. By embedding modelling and decision-support capabilities within shared operational environments, Hybrid EOC’s facilitate continuous interaction between analysts and decision-makers. This proximity enables rapid clarification of assumptions, iterative refinement of scenarios, and collective interpretation of uncertainty, thereby supporting more resilient decision processes (FEMA, 2017; ISO, 2018; Turoff et al., 2004).

Large-scale exercises further reinforce these capabilities by exposing decision-makers to realistic stressors in controlled settings. Exercises that deliberately incorporate uncertainty, competing priorities, and time pressure allow organizations to practice adaptive decision-making and to evaluate the usability of modelling and decision-support tools under stress. Insights gained from such exercises inform improvements in both technical systems and management practices.

Decision-making under uncertainty and stress represents a critical interface between advanced modelling and emergency management. Subsequent research in human–automation interaction reinforces this view, showing that over-reliance on automated or highly optimized decision-support can undermine situational awareness and accountability if human authority and interpretative control are not explicitly preserved (Endsley, 2017). Technical sophistication alone cannot resolve the cognitive and organizational challenges inherent in crisis environments; effective decision-making depends on integrated approaches that align analytical tools, human judgement, and organizational structures.

From a complex adaptive systems perspective, trust emerges as a systemic property shaped by repeated interactions between information quality, communication practices, and decision outcomes.

In this context, ethical considerations arise not from the use of advanced modelling itself, but from issues of transparency, accountability, responsibility, and trust when analytical outputs inform high-stakes decisions under conditions of uncertainty.

4.7 Information management, communication, and trust

Information management and communication constitute critical enablers of effective nuclear and radiological emergency preparedness and response. Advanced modelling and decision-support systems generate essential analytical insights, yet their operational impact depends on how information is organized, communicated, and trusted within and across institutions, as well as by external stakeholders. In environments characterized by uncertainty, high public sensitivity, and rapid information dissemination, deficiencies in information management can undermine decision-making even when technical assessments are robust.

Information management in emergency contexts extends beyond data collection and storage. It involves the continuous integration, validation, and prioritization of heterogeneous information streams, including monitoring data, modelling outputs, field reports, expert assessments, and situational updates from multiple organizations. As incidents evolve, maintaining coherence and traceability across these streams becomes increasingly challenging. Without structured information architectures, inconsistencies, duplication, and information overload can erode situational awareness and delay decisions (Bankes, 1993; Deb, 2001; Walker et al., 2013; Lempert, 2019).

Kavan et al. (2025) and Renn (2008) advanced modelling introduces additional complexity into information management processes. Modelling outputs are inherently scenario-dependent and accompanied by assumptions and uncertainty ranges that may not be immediately intuitive to non-technical audiences. When multiple models or analytical teams are involved, divergent outputs may emerge, further complicating interpretation. Effective information management therefore emphasizes synthesis and contextualization, ensuring that modelling results are framed in relation to operational objectives, decision timelines, and available response options.

Communication represents the interface through which analytical insights are translated into coordinated action and public understanding. Internally, communication among technical experts, emergency managers, and decision-makers must support iterative clarification of assumptions, uncertainties, and implications. Externally, communication with the public, media, and international partners must balance transparency with clarity, avoiding both unnecessary alarm and unwarranted reassurance. The way modelling outputs inform communication strategies can significantly influence public trust and compliance with protective actions (Renn and Levine, 1991; Covello, 2003; World Health Organization, 2017).

Trust plays a central role in mediating the effectiveness of information and communication processes. From a governance perspective, trust in emergency institutions is increasingly recognized as a critical determinant of public compliance and response effectiveness, shaped by transparency, consistency, and perceived competence in decision-making under uncertainty (OECD, 2020).

Within response organizations, trust in modelling and decision-support systems affects whether analytical outputs are accepted, questioned, or disregarded. This trust is shaped not only by technical performance, but also by transparency, explainability, and prior experience. Organizations that routinely engage with modelling tools through training and exercises are more likely to trust and effectively use these tools during crises.

Public trust presents additional challenges. Discrepancies between communicated messages and subsequent outcomes, particularly when not adequately explained, can undermine credibility and fuel misinformation. Conversely, transparent acknowledgement of uncertainty when communicated in a structured and comprehensible manner—can enhance trust by demonstrating competence and honesty. Achieving this balance requires close collaboration between technical experts and communication specialists.

Hybrid operational frameworks provide structural support for improved information management and communication. By integrating modelling outputs, situational data, and communication tools within shared operational environments, Hybrid Emergency Operations Centers facilitate consistent messaging and coordinated information release. Common visualization platforms and shared situational dashboards reduce the risk of divergent narratives and support alignment across institutions and jurisdictions.

The rapid dissemination of information through digital media further amplifies the importance of coordinated communication. Misinformation and disinformation can spread quickly, complicating response efforts and eroding trust even in the presence of sound technical analysis. Addressing these challenges requires proactive communication strategies that are informed by modelling outputs, synchronized across institutions, and responsive to public concerns.

Information management, communication, and trust form an interconnected triad that shapes the operational relevance of advanced modelling and decision-support systems. Modelling enhances situational understanding, but its value is realized only when information is coherently managed, effectively communicated, and trusted by decision-makers and the public alike (French et al., 2009; Endsley, 1995; Parasuraman and Riley, 1997).

Figure 5 Information management, communication, and trust as an interconnected triad shaping the operational relevance of advanced modelling and decision-support systems. Advanced modelling enhances situational understanding, but its operational value is realized only when information is coherently managed, effectively communicated, and trusted by decision-makers and the public. Within complex adaptive emergency management systems, this triad conditions how analytical outputs are interpreted and translated into coordinated action.

4.8 Synthesis of section 4 and transition to discussion

The preceding section examined advanced modelling and decision-support in nuclear and radiological emergency preparedness and response lenses through the lens of complex adaptive systems. Taken together, they illustrate a fundamental shift in where the principal challenges of emergency management now reside. While analytical capabilities have advanced substantially over recent decades, the limiting factors for effective response increasingly lie in integration, interpretation, governance, and human decision-making rather than in computational performance alone.

The evolution of dispersion modelling discussed in Section 4.1 demonstrates that these tools were originally conceived to support preparedness activities scenario exploration, planning, zoning, and exercises rather than providing prescriptive guidance during real-time response. Advances in model fidelity, integration, and computational speed have strengthened preparedness and expanded analytical insight, but they have not altered the foundational role of modelling as an exploratory, rather than directive, instrument. This historical perspective is essential for understanding why tensions arise when preparedness-oriented outputs are implicitly expected to guide operational decisions under uncertainty.

Section 4.2 extended this argument to decision-support systems, showing how DSS architectures have largely inherited the logic of preparedness. Scenario-driven workflows, predefined thresholds, and procedural pathways have contributed to consistency, traceability, and regulatory compliance. At the same time, these features can constrain flexibility in dynamic response environments where situations evolve beyond anticipated scenarios. Within a complex adaptive system, decision-making cannot be reduced to the execution of predefined logic, but emerges from continuous reasoning about trade-offs, timing, feasibility, and coordination among multiple actors.

The introduction of Hybrid EOC in Section 4.3 represents a response to these limitations, not by replacing modelling or DSS, but by repositioning them within an integrative governance framework. Hybrid EOCs function as boundary-spanning environments that connect analytical tools, organizational processes, and communication practices. Rather than seeking to eliminate uncertainty or centralized control, they enable coordinated adaptation by supporting shared interpretation, transparency, and accountability across institutional boundaries. In this sense, Hybrid EOCs operationalize the recognition that decision-making in emergencies is a socio-technical process rather than a purely analytical one.

Section 4.4 further reinforced this perspective by examining the applicability of existing DSS to compound and cascading emergencies. Such events expose the limits of single-hazard assumptions and radiological-centric decision logic, particularly when multiple hazards interact with infrastructure failures, resource constraints, and organizational disruptions. While DSS remain indispensable for providing domain-specific insight, their effectiveness in compound emergencies depends on their embedding within broader coordination mechanisms capable of accommodating interacting hazards and competing priorities.

Sections 4.5, 4.6 highlighted the role of scenario-based planning, large-scale exercises, and decision-making under uncertainty and stress as critical mechanisms for bridging analytical capability and operational reality. Exercises serve not only to validate technical systems, but to expose how modelling outputs, decision-support tools, organizational processes, and human judgement interact under realistic conditions. They reveal that many challenges arise not from insufficient analytical accuracy, but from misalignment between information flows, decision timelines, cognitive demands, and institutional responsibilities.

Across these sections, a consistent theme emerges advanced modelling and decision-support systems derive their operational value not from their ability to generate increasingly precise outputs, but from how effectively they are integrated into human-centered, governance-aware decision environments. Uncertainty, stress, and complexity are not anomalies to be eliminated, but inherent features of nuclear and radiological emergencies that must be managed through adaptive reasoning, coordination, and trust.

This synthesis provides the foundation for discussion in Section 5. The following section moves beyond descriptive analysis to examine the implications of these findings for future research, system design, and operational practice. It explores how emerging approaches including adaptive decision exploration, hybrid operational concepts, and advanced computational methods can be systematically developed, validated, and operationalized without undermining accountability or human judgement. In doing so, the Discussion addresses how nuclear and radiological emergency preparedness and response can evolve from technically sophisticated yet fragmented systems toward integrated, resilient, and adaptive response ecosystems.

Figure 6 synthesizes the progression from preparedness-oriented modelling and scenario-driven decision-support systems toward integrated governance frameworks enabled by Hybrid EOC. Rather than prescribing actions, advanced analytical tools support adaptive, human-in-the-loop decision exploration, in which trade-offs, uncertainty, and organizational constraints are explicitly considered within complex adaptive emergency management systems.

5 Discussion

Some of the conceptual perspectives discussed in this review were built upon earlier doctoral research by the author on systemic modelling and exploratory optimization approaches in nuclear emergency preparedness (Min, 2020).

The arguments developed in this review should not be interpreted as a critique of modelling rigor or analytical advancement, but as a reflection on how such capabilities are governed, interpreted, and integrated into institutional decision-making under conditions of uncertainty and complexity.

Building on the dual analytical perspective developed in previous section, distinguishing between prevailing theoretical paradigms in modelling and decision-support and the applied and conceptual insights derived from their operational limitations, this discussion examines the implications of these findings for EPR.

Rather than reiterating technical advances, the discussion focuses on how modelling and decision-support capabilities are embedded within organizational, institutional, and governance contexts, and how these contexts shape their effectiveness under conditions of uncertainty, time pressure, and multi-actor coordination. This perspective enables a transition from analytical description toward reflection on adaptive management, human-centered decision-making, and future research directions.

5.1 From deterministic decision support to adaptive decision exploration

From an applied perspective, the limitations identified in theoretical modelling and decision-support paradigms highlight the need for governance arrangements that explicitly accommodate uncertainty, organizational interaction, and adaptive coordination.

The analysis presented in the previous sections highlights a persistent tension in EPR between analytical sophistication and operational decision-making. Over time, modelling and decision-support systems have become increasingly advanced, providing detailed projections of hazard evolution and potential consequences. However, real-world emergencies repeatedly demonstrate that improved analytical precision does not automatically translate into better decisions.

Traditional decision-support approaches in EPR have largely been shaped by deterministic or scenario-based paradigms. These paradigms assume relatively stable system behavior and predefined response pathways, often encoded through thresholds, procedures, and standardized decision trees. While such approaches support consistency and regulatory compliance, they may constrain adaptive reasoning when emergencies evolve in unexpected ways or when multiple objectives and constraints compete (IAEA, 2021; OECD Nuclear Energy Agency (NEA), 2014; Comfort et al., 2010).

In complex and rapidly evolving situations, decision-makers are rarely confronted with a single “correct” course of action. Instead, they must navigate a space of plausible, feasible, and socially acceptable response strategies, each associated with different trade-offs. This perspective aligns with more recent work on decision-making under deep uncertainty, which emphasizes robustness, exploration of alternative futures, and the evaluation of trade-offs across a wide range of plausible assumptions rather than reliance on optimal solutions derived from a single predictive model (Kwakkel et al., 2020).

From this perspective, the central limitation of deterministic decision-support systems is not their analytical accuracy, but their limited capacity to support adaptive reasoning by human decision-makers under uncertainty and time pressure.

This observation motivates a shift from decision-support systems designed primarily to recommend actions toward frameworks that support decision exploration. In such frameworks, the role of analytical tools is to structure the decision space, reveal trade-offs, and inform judgement, rather than to prescribe optimal solutions.

5.2 Human-in-the-loop decision-making and the governance of complexity

The emphasis on human-in-the-loop decision-making emerges directly from the review of modelling and decision-support paradigms, which consistently demonstrate that analytical tools cannot replace institutional responsibility, contextual judgement, or ethical accountability.

Conceptualizing EPR as a complex adaptive system has profound implications for decision-making governance. In such systems, outcomes emerge from interactions among technical, organizational, human, and societal subsystems, rather than from linear cause–effect relationships. As a result, attempts to fully automate decision-making or to rely exclusively on algorithmic optimization risk undermining institutional accountability, transparency, and trust.

A human-in-the-loop paradigm is therefore not a technical limitation, but a governance requirement in emergency management. Human decision-makers contribute contextual understanding, ethical judgement, and responsibility that cannot be meaningfully reduced to algorithmic rules. Decision-support tools must be designed to augment these capabilities, not to replace them.

Within this context, the Hybrid EOC can be interpreted as a mechanism for governing complexity. By integrating modelling outputs, decision-support analytics, organizational workflows, and communication processes, Hybrid EOCs provide a structured environment in which institutions can coordinate actions, align interpretations, and maintain traceability under uncertainty. Importantly, such environments do not seek to eliminate uncertainty or enforce convergence toward a single optimal decision, but to enable coherent institutional action despite uncertainty.

This governance perspective becomes particularly relevant in multi-actor and compound emergency scenarios, where nuclear or radiological incidents interact with other hazards, infrastructure disruptions, or cross-border dynamics. In such cases, effective response depends as much on coordination, prioritization, and trust as on technical accuracy.

Figure 7 illustrates the human-in-the-loop paradigm, in which analytical tools, organizational frameworks, and societal considerations support, rather than replace human institutional decision authority. Within complex adaptive emergency management systems, this layered structure preserves accountability and legitimacy while enabling informed and adaptive decision-making under uncertainty (Klein, 1998; Kahneman, 2011; Flin et al., 2008).

5.3 Emerging research directions: adaptive computation in EPR

Building on gaps identified across the reviewed literature, advanced computational approaches such as evolutionary and learning-based methods are discussed here not as operational decision-makers, but as preparedness-oriented tools for decision exploration and organizational learning.

Within the context of EPR as a complex adaptive system, advanced computational approaches such as evolutionary and learning-based methods may offer promising avenues for future research. Their relevance lies not in automating decisions or identifying single optimal responses, but in supporting the structured exploration of alternative response strategies and associated trade-offs under uncertainty.

When embedded within human-in-the-loop decision-support frameworks and governed by appropriate transparency and accountability mechanisms, such approaches could contribute to enhanced decision intelligence. However, their practical value depends critically on how they are integrated into organizational processes and governance structures, rather than on algorithmic performance alone.

From a research perspective, these approaches are best understood as tools for exploring decision spaces rather than optimizing outcomes. Their potential lies in generating diverse sets of feasible strategies, examining robustness across varying assumptions, and supporting reflective decision-making under pressure. Importantly, this perspective aligns with the understanding of uncertainty as an intrinsic property of complex systems, rather than as a deficit to be eliminated.

At the same time, the introduction of advanced computational methods raises important ethical and institutional questions. Transparency, explainability, auditability, and preservation of human authority must be treated as core design constraints. Without these safeguards, the use of adaptive computation risks reinforcing automation bias or eroding accountability, particularly in high-stakes emergency contexts.

An important and often underexplored opportunity for advanced computational approaches lies in the preparedness phase rather than in real-time response. In this phase, evolutionary and learning-based methods can be used to systematically explore response arrangements, test their robustness across a wide range of scenarios, and support institutional learning without the pressures and risks associated with operational decision-making.

By operating in an offline, controlled environment, such approaches can contribute to the development of pre-agreed response arrangements and contingency options that enhance preparedness while preserving human authority during response. In this sense, advanced computation supports preparedness as a process of structured learning and adaptation, rather than as a mechanism for automating emergency decisions.

Figure 8 illustrates the conceptual distinction between preparedness and response phases in nuclear and radiological EPR. Advanced computational approaches support preparedness through scenario exploration, modelling, and learning, while response is governed by human-centered decision-making, institutional coordination, and accountability within hybrid operational frameworks.

5.4 Implications for practice and future research

The discussion above suggests that improving EPR is not primarily a matter of further increasing model fidelity or computational power. Rather, it requires a rethinking of how analytical capabilities are embedded within decision-making and governance frameworks.

For practice, this implies a greater emphasis on integrative environments that enable coordination across technical, organizational, and societal domains. Existing modelling and decision-support tools remain indispensable, but their effectiveness depends on how they are contextualized, interpreted, and used within broader emergency management processes, particularly in complex and compound emergencies (Pescaroli and Alexander, 2015; Gill and Malamud, 2014; OECD, 2018).

For research, these observations point to the need for interdisciplinary approaches that bridge modelling, decision-support, organizational studies, and governance. Evaluating decision-support capabilities in realistic operational contexts such as exercises and simulations will be essential for understanding their impact on decision quality, coordination, and trust.

5.5 Concluding perspectives and research-to-practice transition

This review has examined advanced modelling and management strategies for nuclear and radiological incidents through the lens of complex adaptive systems. The analysis highlights the limitations of deterministic, prescriptive decision-support approaches and underscores the importance of adaptive, human-centered frameworks capable of operating under uncertainty and complexity.

A key implication is that future progress in EPR is likely to depend less on isolated technical advances and more on coordinated research and innovation efforts that integrate modelling, decision-support, and organizational governance. Such efforts are necessary to translate conceptual advances into operational capabilities validated in realistic emergency contexts.

By explicitly linking complex systems theory, decision-support research, and institutional practice, this perspective points toward a new generation of EPR capabilities adaptive, transparent, and resilient that strengthen both technical performance and societal trust in emergency decision-making.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The author acknowledges the use of generative artificial intelligence tools for figure ideation and language editing during manuscript preparation.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Nadja Zeleznik, Milan Vidmar Electric Power Research Institute, Slovenia

Reviewed by: Stephen Musolino, RadSafe Solutions LLC, United States

Stepan Kavan, University of South Bohemia in České Budějovice, Czechia