Health Security inequalities in Non-EU European Countries: A Cross-National Comparative Assessment Using an Integrated MCDM-Machine Learning Approach

2.1 Health security indicator compilation

This study utilizes 2019 and 2021 GHSI datasets to assess health security capacities across six domains:²⁰ (Global Health Security Index, Global Health Security Index Data Model and Report (2021)).

Available at (https://ghsindex.org/report-model/). The Global Health Security Index (GHSI) was selected for its established role in benchmarking global pandemic preparedness.

To ensure consistency and comparability between the two time periods, this study employs the revised 2019 GHSI dataset, released alongside the 2021 GHSI report, which accounts for minor methodological adjustments made between the two editions.

This study focuses on 27 non-EU European countries: Albania, Andorra, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Georgia, Iceland, Israel, Kazakhstan, Kyrgyz Republic, Liechtenstein, Moldova, Monaco, Montenegro, North Macedonia, Norway, Russia, San Marino, Serbia, Switzerland, Tajikistan, Turkey, Turkmenistan, Ukraine, the United Kingdom, and Uzbekistan. These nations, while geographically and culturally connected to Europe to varying degrees, operate outside the EU’s political-economic framework, presenting unique health security governance challenges and opportunities.

Nine performance metrics for six Global Health Security Index (GHSI) indicators (2019, 2021, and aggregated 2017–2021 data) were extracted for non-EU European countries, African, and Eastern Mediterranean (EMR) regions from the Global Health Security Index.²⁰ The non-EU metrics were evaluated against published analyses of Africa²¹ and the EMR,²² enabling cross-regional entropy-based comparisons of health security priorities, disparities, and temporal shifts.

For dynamic cross-national rankings and clustering, the non-EU data were analyzed using an integrated Entropy-CoCoSo (Combined Compromise Solution) and K-means model, categorizing countries into five risk clusters (high to dangerous). Trends, shared strengths, and weaknesses among clustered non-EU countries were assessed by analysing average scores across all 34 sub-indicators (2017–2021).

Comprehensive data are provided in Tables 1–11 (Sheet 1) of the supplementary software (MCDM-ML Based Tool for Health Security Optimization in Non-EU Countries Comprehensive Assessment and Visualization Workflow.xlsm.) as well as in the first section of the supplementary dataset file (The data and materials supporting the results and analyses of the paper.pdf ).⁴³ Source code and datasets are available at [https://github.com/ProfAdelAbdulsalam/Supplementary-material--Software/tree/V2.0.1] and archived via [https://doi.org/10.5281/zenodo.15185666].⁴³

2.2 Assessing cross-regional variations in health security determinants: Entropy-driven estimation of criterion weights

The entropy method, rooted in information theory.⁴⁴ As a foundational approach in MCDM, it assigns objective weights to criteria, minimizing subjective bias in decision analysis. By measuring the uncertainty or “informational value” within a decision matrix, it quantifies data variability to assign higher weights to attributes with greater divergence, ensuring decisions reflect empirical patterns rather than subjective biases.⁴⁵ This data-driven approach eliminates reliance on expert opinions, making it particularly valuable in scenarios where impartiality is critical or expert input is unavailable.⁴⁶ Its versatility extends to handling diverse data types, including exact values and intervals, enhancing its applicability across fields like resource allocation, healthcare, and health security assessment.²¹ While the entropy method’s effectiveness hinges on data quality, its adaptability and theoretical foundation solidify its role as a cornerstone in modern MCDM systems, bridging algorithmic precision with practical health security decision-making demands.^21,22

The entropy method in this study follows a systematic five-step framework to derive objective criterion weights, progressing from data organization to weight computation. Below is a detailed breakdown^21,44:

The process begins by structuring a decision matrix (G), as defined in Equation (1), where rows represent m countries and columns correspond to n health security indicators. Each element ${\mathrm{g}}_{\mathrm{ij}}$ denotes the performance score of country i for indicator j:

For the j-th indicator, the entropy measure is calculated using Equation (3) to quantify its information content within the dataset, where lower entropy signifies higher informational utility.

The relative importance of each criterion is determined by its divergence $(1-{\mathrm{Ent}}_{\mathrm{j}})$ , which is normalized to compute final weights ${\mathrm{W}}_{\mathrm{j}}$ ( Equation 4). Indicators with greater variability receive higher weights:

The research expanded its examination of Health Security (HeS) indicators beyond non-European Union countries, with a particular focus on Africa and the Eastern Mediterranean Region (EMR). Employing the entropy method across nine iterations (encompassing 3 regions and 3 time periods: 2019, 2021, and 2017–2019), the study identified region-specific factors that impact health security. This methodological approach ensured consistency, facilitated cross-regional comparisons, and supported the development of targeted policy recommendations while maintaining global applicability.

Detailed calculations for this step are readily available in (Sheet 2.1 (a, b, c), Sheet 2.2 (a, b, c), and Sheet 2.3 (a, b, c)) of the supplementary software (MCDM-ML Based Tool for Health Security Optimization in Non-EU Countries Comprehensive Assessment and Visualization Workflow.xlsm.).⁴³ Source code and datasets are available at [https://github.com/ProfAdelAbdulsalam/Supplementary-material--Software/tree/V2.0.1] and archived via [https://doi.org/10.5281/zenodo.15185666].⁴³

2.3 Prioritizing Non-EU countries using an integrated entropy-CoCoSo method

This study employs the Combined Compromise Solution (CoCoSo) method, a robust multi-criteria decision-making (MCDM) technique, to evaluate and rank non-EU countries based on a set of predefined criteria. The CoCoSo method’s strength lies in its ability to integrate three aggregation strategies – weighted sum model (WSM), weighted product model (WPM), and an exponential compromise strategy – to derive a balanced ranking of alternatives.⁴⁷ By synthesizing outcomes from distinct mathematical models, this approach ensures robustness and reduces the bias inherent in single-aggregation techniques.⁴⁶

In addition to its effective application in evaluating health security practices, as highlighted in Study,²³ numerous studies have demonstrated the efficiency of integrating methods like entropy in various domains. For instance, Research³³ evaluated the progress of Indian Union Territories (UTs) toward achieving Sustainable Development Goals (SDGs) using the SDG India Index 3.0 (2020–2022 dataset). To tackle the contextual variability inherent in sustainability frameworks, the authors employed a hybrid multicriteria decision-making (MCDM) model that combined Shannon Entropy—leveraging data variability to objectively weight SDG indicators—and the COCOSO (Combined Compromise Solution) method, which ranked UTs by balancing multiple compromise scores.

The integrated Shannon Entropy-COCOSO framework effectively quantified disparities in SDG implementation across the Indian Union Territories. The entropy weighting mechanism allowed for the objective prioritization of indicators based on their variability, while the COCOSO method facilitated the ranking of regions through comprehensive multi-criteria compromise scores. This analysis revealed notable extremes, with Chandigarh ranked highest and Dadra & Nagar Haveli/Daman & Diu positioned lowest in terms of SDG achievement. The model’s robust validation highlights its utility in identifying context-specific gaps, thereby guiding targeted interventions aimed at aligning underperforming regions with Agenda 2030.

Similarly, Study⁴⁸ developed collagen-based packaging films (CSP and CSC) from leather and food waste and evaluated their performance through a hybrid Entropy-CoCoSo framework. This framework utilized entropy weighting to objectively prioritize material properties, such as water solubility and biodegradability, while CoCoSO aggregated multi-criteria scores to rank disparities between different film formulations.

These examples underscore the powerful advantages of integrating methodologies, as they not only enhance analytical robustness but also provide nuanced insights that are essential for formulating effective, targeted strategies in diverse fields.

In this study, we implemented the CoCoSo algorithm in four stages, assigning criteria weights using the entropy method to ensure objectivity, as validated in recent MCDM frameworks.^33,48 The CoCoSo method was chosen for its unique ability to handle trade-offs between conflicting criteria while providing stable and interpretable rankings, as demonstrated by its superior performance in comparative analyses compared to other MCDM methods.^35,46 The key steps of the CoCoSo method are outlined as follows⁴⁷:

The method begins with the decision matrix G = [ ${\mathrm{g}}_{\mathrm{ij}}$ ], where ${\mathrm{g}}_{\mathrm{ij}}$ represents the performance score of alternative i for criterion j, as shown in Equation (1). Normalization is applied to eliminate scale differences using Equation (5):

Here, ${\mathrm{g}}_{\mathrm{j}}^{+}$ and ${\mathrm{g}}_{\mathrm{j}}^{-}$ denote the maximum and minimum values of criterion j, respectively, across all countries i.

Additive Weighted Sequence ( Equation 6):

Exponential Weighted Sequence ( Equation 7):

Here, ${\mathrm{w}}_{\mathrm{j}}$ is the weight of criterion j, derived using the entropy method to ensure objectivity.

Three appraisal scores are calculated to evaluate alternatives: Normalized Arithmetic Mean ( Equation 8); Relative Superiority Score ( Equation 9); and Balanced Compromise Score ( Equation 10).

The parameter λ (0 ≤ λ ≤ 1) allows decision-makers to adjust the influence of S_i (additive) and P_i (exponential). A default λ = 0.5 ensures equal weighting.

The final composite score C_i ( Equation 11) combines all three appraisal scores:

Alternatives are ranked in descending order of ${\mathrm{C}}_{\mathrm{i}}$ , with higher values indicating superior performance.

This study applied the entropy-CoCoSo algorithm to rank Non-EU countries across three time periods (2019, 2021, 2017–2019), evaluating preparedness and capabilities for addressing global health challenges.

Detailed calculations for this step are readily available in (Sheet 3 (a, b, c)) of the supplementary software (MCDM-ML Based Tool for Health Security Optimization in Non-EU Countries Comprehensive Assessment and Visualization Workflow.xlsm.).⁴³ Source code and datasets are available at [https://github.com/ProfAdelAbdulsalam/Supplementary-material--Software/tree/V2.0.1] and archived via [https://doi.org/10.5281/zenodo.15185666].⁴³

2.4 Clustering

K-means clustering is a centroid-based machine learning method that categorizes unlabelled data into a specified number of clusters by iteratively optimizing the proximity of data points to cluster centroids.³⁶ By grouping observations into homogeneous clusters, this algorithm maximizes similarity within each group while minimizing variance between them.

The process begins by initializing K cluster centroids, with each data point assigned to the nearest centroid using distance metrics such as Euclidean distance, forming provisional clusters. Subsequently, the centroids are recalculated as the mean of all points within their respective clusters, and this iterative cycle of assignment and centroid updates continues until the centroids stabilize (with minimal positional change) or a maximum iteration threshold is reached.³⁶

Notably, building upon the foundational work highlighted in previous studies and its successful application in prior health security analyses to classify countries in various regions,^21–23 K-means clustering’s simplicity and computational efficiency make it well-suited for this analysis.²¹ By categorizing countries into homogeneous clusters based on key health security indicators, we can effectively discern patterns and disparities that may not be immediately evident through other analytical methods.²² Specifically, we utilized K-means clustering to partition the data into stratified groups, allowing for a nuanced examination of varying health security profiles.²³ This approach enables us to visualize and contextualize the differences between countries, facilitating a targeted analysis of regional vulnerabilities.^21–23

Furthermore, by employing clustering, we can track shifts in health security performance over time, providing valuable insights for policymakers to implement tailored interventions.^21–23 In combining this method with advanced multi-criteria decision-making techniques—such as entropy weighting and the CoCoSo method—we reinforce the robustness of our findings. The integration of clustering exemplifies a methodological synergy that not only enhances our analytical rigor but also empowers stakeholders with actionable insights derived from complex data landscapes.^21–23

In our study, we adopted a five-level categorization framework (1–5) to evaluate regional performance across countries, with a rating scale spanning from 1 (high) to 5 (dangerous), and intermediate tiers labelled as 2 (safe), 3 (moderate), and 4 (alert), allowing for a granular assessment of regional disparities. To classify non-EU nations, we repeated the clustering procedure across three independent iterations, each corresponding to outcomes derived from a composite ranking methodology applied to three distinct datasets.

To facilitate the clustering process of Non-EU countries across three time periods (2019, 2021, 2017–2019), we utilized an open source software,⁴⁹ a freely available tool for segmentation tasks. (https://www.clusteranalysis4marketing.com/a-marketers-guide-to-cluster-analysis/free-download/). Then the clustering results were integrated into our developed software for further analysing process, detailed calculations for this step are readily available in (Sheet 4 (a, b, c)) of the supplementary software (MCDM-ML Based Tool for Health Security Optimization in Non-EU Countries Comprehensive Assessment and Visualization Workflow.xlsm.).⁴³ Source code and datasets are available at [https://github.com/ProfAdelAbdulsalam/Supplementary-material--Software/tree/V2.0.1] and archived via [https://doi.org/10.5281/zenodo.15185666].⁴³

2.5 Methods justification and statistical validation

The choice of the integrated Entropy–CoCoSo–K-means framework was motivated by its complementary strengths and capacity to tackle the complexity of multidimensional health security assessment. Each method was selected for specific, mutually reinforcing reasons directly related to the workflow in Sections 2.1–2.4.

1. Entropy Method – Chosen for its objectivity and data-driven weighting capability (Section 2.2). It quantifies the informational value of each health security indicator, assigning higher weights to those with greater variability across countries and periods. This approach eliminates the reliance on expert judgment, thereby minimizing subjective bias, which is particularly crucial in international comparisons where standardized expert input may not be available.

2. CoCoSo Method – Selected for its robust ranking ability and integration of multiple aggregation strategies (Section 2.3). By combining the weighted sum model, weighted product model, and exponential compromise approach, CoCoSo balances trade-offs among conflicting indicators, producing rankings that are stable, interpretable, and resilient to data variability.

3. K-means Clustering – Employed for its computational efficiency and proven effectiveness in pattern discovery (Section 2.4). K-means groups countries into homogeneous clusters based on performance scores, revealing structural patterns and disparities that may be obscured in the raw rankings. This facilitates policy-relevant categorization and enables the tracking of temporal performance.

The integrated use of these three methods has been successfully applied in recent EU-focused research on health security, demonstrating that this framework can objectively identify priority dimensions, produce robust country rankings, and classify nations into meaningful performance tiers, yielding actionable policy insights in a complex regional setting.⁴² Building on this established effectiveness in the EU context, our study extends this approach to non-EU European countries, enabling cross-national comparability while accounting for regional differences in governance, resources, and public health infrastructure.

To ensure the statistical validity of our clustering results, we used the silhouette score, which is a widely used metric for assessing cluster cohesion and separation.^50,51 Our five-cluster solution achieved a silhouette score of 0.57, reflecting a reasonable degree of separation between the clusters and cohesion within them. This score confirms that the clusters are well-defined rather than arbitrary, providing statistical support for our categorization of countries into distinct performance tiers. This validation step reinforced the reliability of our findings and underscored the suitability of the K-means algorithm for this analysis.

2.6 Software

For this study, a simple Excel-based software tool was developed to systematically assess Health Security (HeS) inequalities across non-EU European countries.⁴³

Source code and datasets are available at [https://github.com/ProfAdelAbdulsalam/Supplementary-material--Software/tree/V2.0.1] and archived via [https://doi.org/10.5281/zenodo.15185666].⁴³

The Main Menu of the software tool serves as the central navigation interface, guiding users through the sequential workflow of the integrated Multi-Criteria Decision-Making (MCDM) - Machine Learning methodology. It illustrates the Four-step implementation and analytical process described in this research.

For Visualization, this tool provides access to tabular and graphical outputs, including entropy weights, rankings, and cluster patterns. In addition, the tool’s structured layout mirrors the research methodology, enabling seamless navigation between data input, computational analysis, and result interpretation. Interactive hyperlinks to corresponding sheets or modules ensure a user-friendly experience while maintaining alignment with the paper’s technical framework.

3.1 Entropy-driven assessment of HeS priorities

Table 1 presents the entropy-weighted prioritization of health security dimensions across regions. Weights for six GHSI indicators in non-EU Europe, Africa, and the EMR (2019, 2021, aggregated 2017–2021), It compares the relative weights of six Global Health Security Index (GHSI) indicators. Together with Figure 2, this table offers critical insights into regional prioritization patterns of HeS indicators. Figure 2 illustrates the comparative distribution of weights across HeS dimensions for non-EU countries, the African Region, and the EMR over three timeframes: 2019, 2021, and the aggregated period of 2017–2021. Complementing this analysis, Table 2 illustrates the comparative entropy weight changes (2019–2021) across regions. Positive/negative values indicate increased/decreased prioritization of indicators over time.

Figure 1. Methodological Flowchart.

A flowchart illustrating the four-stage analytical process: (1) Data Compilation from GHSI, (2) Criterion Weighting using the Entropy Method, (3) Country Prioritization using the CoCoSo Method, and (4) Performance Clustering using K-means.

3.2 Health security rankings and cluster-based interventions

Table 3 outlines the performance metrics and comparative standings of non-EU countries derived from the Entropy-CoCoSo HeS methodology. For three intervals—2019, 2021, and the cumulative 2017–2021 timeframe—the table displays computed scores (Ci) and their ordinal positions (Ri). Each country is further classified into one of five performance tiers (Si), where S1denotes optimal outcomes (‘High Performance’) and S5signals critical deficiencies (‘Dangerous Level’). The data also track fluctuations in rankings and tier placements between 2019 and 2021, illustrating shifts in national progress or regression. Figure 3 visually maps the ranked outcomes, while Figure 4 depicts the spatial distribution of clusters across this group of countries.

Figure 2. Entropy-weighted prioritization of health security dimensions across non-EU countries, EMR, and African Region, 2017–2021 (Source: Authors).

This figure illustrates the entropy-weighted prioritization of health security dimensions across non-EU European countries, the Eastern Mediterranean Region (EMR), and the African Region (2017–2021). The figure compares the relative weights of six Global Health Security Index (GHSI) indicators—Prevention, Detection & Reporting, Rapid Response, Health System, Compliance with Norms, and Risk Environment—across regions and timeframes.

Figure 3. Health Security Preparedness Rankings of Non-EU countries (Source: Authors).

This bar chart illustrates the Health Security Preparedness Rankings of non-EU European countries, encompassing the years 2019, 2021, and the entire period of 2017-2021. Countries are ranked based on their Composite CoCoSo scores, with higher values indicating superior performance. It also reflects temporal shifts in health security capabilities.

4.1 Comparative weighting of health security indicators: Non-EU, Africa, and EMR perspectives

4.1.1 Discussion of Non-EU Countries’ health security priorities

The health security landscape among non-European countries reveals significant trends and variations in prioritization, as evident in Table 1 and Figure 2. A notable finding is the consistent emergence of detection & reporting as the most influential HeS indicator, with weights ranging from 0.372 to 0.399 across different time periods. This underscores the critical importance of surveillance capabilities and data transparency, which are essential components in identifying and managing public health threats. The slight decline in its weight by 2021, from 0.399 to 0.372, may reflect challenges in `maintaining robust reporting mechanisms amidst the complexities posed by emerging health crises, such as the COVID-19 pandemic, which exposed weaknesses in global surveillance systems.⁵² This decline highlights the need for sustained efforts to strengthen surveillance capabilities.

The increasing weight of prevention, from 0.204 to 0.218 over time, signifies a growing recognition of the need for pre-emptive measures, such as vaccination and overall public health preparedness. This shift resonates with global trends that emphasize the necessity of robust preventive strategies to mitigate the impact of future health emergencies, as emphasized by WHO initiatives aimed at enhancing vaccine coverage globally.⁵³ However, the disparities observed in vaccination rates and public health preparedness among non-EU countries hint at persistent inequities that require concerted efforts to address.

In contrast, compliance with international norms shows a significant increase from 0.091 to 0.123 post-2019, indicating emerging regulatory gaps or uneven adherence to established health standards. This trend may reflect the ongoing developments in international health regulations, such as the International Health Regulations (IHR), and recent reforms in Europe, which have urged nations to align their practices with global and regional standards.⁵⁴ The rising focus on compliance underscores the necessity for stricter regulatory enforcement to curb health security vulnerabilities that could compromise global health outcomes.

Furthermore, indicators such as Rapid Response (0.097–0.104) and Health System (0.126–0.135) are less influential, suggesting systemic weaknesses in emergency coordination and healthcare infrastructure across these countries. The relative stagnation of these areas points to the challenges faced by non-EU states in developing swift and effective responses to health emergencies, a concern that aligns with findings from global health security initiatives highlighting the importance of strengthening health systems in those countries.⁵⁵

Additionally, the Risk Environment indicator remains minimally influential (~0.06–0.067), reflecting persistent vulnerabilities faced by non-EU countries concerning external threats, such as climate change or geopolitical instabilities. This finding parallels discussions in public health literature concerning the interplay between environmental factors and health security, suggesting that a multi-faceted approach that encompasses both health and environmental policies is critical.⁵⁶

The findings indicate that non-EU countries may be prioritizing immediate, quantifiable security concerns at the expense of long-term environmental and geopolitical risks. This narrow focus could render them ill-prepared for complex, interconnected challenges—such as climate change, resource scarcity, and evolving geopolitical dynamics—that transcend traditional security frameworks.⁵⁷ Such a limited perspective on traditional security frameworks may impede the development of comprehensive policies that address the root causes of insecurity and enhance resilience against future challenges. For example, non-EU countries might not fully exploit collaborative opportunities to address global threats due to insufficient attention to long-term geopolitical implications. This oversight could result in missed opportunities for international cooperation and the sharing of resources and knowledge, which are essential for addressing global issues such as climate change and geopolitical instability.⁵⁷ Policymakers should adopt a more comprehensive risk assessment approach that balances short-term priorities with strategic long-term considerations. By integrating environmental, climate, and geopolitical dimensions into national security strategies, resilience can be strengthened, adaptability improved, and preparedness for emerging global threats enhanced. Such a holistic perspective not only reduces vulnerability but also fosters international cooperation in addressing shared risks, ultimately supporting national and regional stability in an increasingly complex global environment.

The aggregated weight stability across these indicators signifies a sustained focus on surveillance and prevention, while the rising weights of compliance underscore the urgent need for enhanced regulatory measures. To address these challenges, policymakers must prioritize the harmonization of detection systems, the scaling up of preventive measures, and addressing compliance gaps to reduce disparities among countries. Moreover, fostering collaboration and knowledge sharing among nations can bolster resilience against emergent health threats, significantly aligning with international efforts to build a more robust global health security framework.⁵⁸ Through such concerted actions, non-EU countries can better navigate the complexities of health security and enhance their overall preparedness for future public health challenges.

4.1.2 Comparative analysis of Non-EU and African health security priorities

Non-EU countries and the African region exhibit stark contrasts in health security prioritization. In non-EU states, detection & reporting dominates (0.372–0.399), reflecting significant disparities in surveillance systems and data-sharing capabilities, while prevention (0.204–0.218) plays a secondary role. Conversely, Africa prioritizes prevention (0.325–0.352) as its top indicator, highlighting variability in pre-emptive measures like vaccine access and sanitation infrastructure, with detection & reporting declining post-2019 (0.326 → 0.294),²¹ signalling converging surveillance challenges. Non-EU’s compliance with norms surged (0.091 → 0.123), pointing to emerging regulatory gaps, whereas Africa’s compliance remains stagnant (~0.05–0.06), suggesting systemic neglect of health standards. Both regions show weak differentiation in Rapid Response (non-EU: 0.097–0.104; Africa: 0.060–0.072), exposing shared deficiencies in emergency coordination.⁵⁹ However, Africa’s health system weights (0.169–0.175) surpass non-EU’s (0.126–0.135), indicating greater disparities in healthcare infrastructure across African nations.⁶⁰ Risk Environment is uniformly low (~0.06), underscoring neglected climate or geopolitical vulnerabilities in both regions. These contrasts necessitate tailored policies: non-EU should harmonize surveillance networks and enforce compliance, while Africa must scale preventive infrastructure and healthcare investments. Cross-regional collaboration—leveraging non-EU’s surveillance expertise to bolster Africa’s detection gaps and Africa’s preventive models to inform non-EU strategies—could address shared weaknesses in crisis response and risk resilience.

4.1.3 Comparative analysis of Non-EU and EMR health security priorities

Non-EU countries and the Eastern Mediterranean Region (EMR) demonstrate divergent health security priorities. Non-EU nations prioritize detection & reporting (0.372–0.399) as the dominant indicator, reflecting variability in surveillance systems, while the EMR focuses overwhelmingly on health systems (0.261–0.322), highlighting disparities in healthcare infrastructure and workforce capacity.²¹ Notably, the EMR’s prevention weight rises sharply (0.209 → 0.259) by 2021, nearing the non-EU’s emphasis (0.204–0.218), suggesting a growing recognition of pre-emptive measures like vaccine equity. However, Detection & Reporting remains far less influential in the EMR (0.217–0.221 vs. non-EU’s 0.372–0.399), indicating weaker differentiation in surveillance capabilities. Both regions share low weights for Rapid Response (non-EU: 0.097–0.104; EMR: 0.092–0.098), underscoring systemic gaps in emergency coordination. Compliance with norms is marginally higher in non-EU (0.091 → 0.123) compared to the EMR’s stagnant values (~0.06–0.063), suggesting non-EU faces emerging regulatory challenges, while the EMR neglects this domain. The EMR also shows slightly higher risk environment weights (~0.095–0.099 vs. non-EU’s ~0.06–0.067), signalling greater variability in political, and socioeconomically vulnerabilities.

4.1.4 Non-EU health security dynamics: Cross-regional contrasts and emerging challenges in global health governance

The examination of non-EU countries’ health security dynamics between 2019 and 2021 reveals significant contrasts and emerging challenges that shape global health governance. Notably, the increased prioritization of prevention (+0.014) and compliance with norms (+0.031) indicates a growing recognition among non-EU nations of the importance of pre-emptive measures and the need to address regulatory gaps. This trend aligns with a global movement towards enhancing preventive health strategies and compliance with international health regulations as essential components of health security.⁵⁴ The rising focus on compliance, in particular, underscores the importance of strengthening regulatory frameworks to ensure adherence to global health standards.

In contrast, the sharp decline in detection and reporting (-0.027) within non-EU countries echoes a similar trend observed in Africa (-0.032). This decline raises concerns about diminishing surveillance capabilities and transparency in reporting health threats. Notably, the Eastern Mediterranean Region (EMR) experienced a slight gain in detection (+0.004), highlighting a divergence in surveillance dynamics that underscores the necessity for tailored approaches in each region. The disparities in detection efforts reflect varying capacities for health monitoring and reporting, necessitating an urgent focus on enhancing the data infrastructure and capabilities across non-EU countries.⁶¹

Furthermore, the loss of relevance in health system indicators (-0.010) among non-EU countries is particularly concerning. This trend diverges from Africa’s marginal decline (-0.004) and the EMR’s steep drop (-0.060), indicating a complex landscape of healthcare vulnerabilities. The systemic weaknesses faced by both non-EU and African nations pose significant challenges to their health security, revealing a pressing need for investment in healthcare infrastructure and systems. Conversely, the worsening infrastructure circumstances in the EMR post-pandemic highlight the urgent need for stabilization efforts, further emphasizing the regional disparities in health system resilience.²²

Interestingly, the rising focus on regulatory compliance in non-EU countries (+0.031) stands in contrast to Africa’s stagnant compliance (-0.006). This finding suggests a fragmented adherence to global health norms, signifying the need for improved mechanisms to ensure that all regions align with international standards. Such contrasts in compliance outcomes align with the assertions of,²² advocating for a more cohesive and collaborative global approach to health governance, capable of bridging gaps and enhancing compliance across diverse contexts.

Despite these emerging improvements in compliance, all regions, including non-EU countries, continue to neglect rapid response capabilities, with non-EU exhibiting a decline of (-0.006). While Africa has shown a slight improvement in this domain (+0.011), the EMR has experienced a minimal drop (-0.003). Africa’s improvement hints at the potential development of emerging crisis coordination efforts, which could be furthered by leveraging gains in prevention and compliance. These insights indicate that while progress is being made, a holistic approach to emergency preparedness remains imperative across all regions.

Ultimately, these contrasting trends highlight the necessity for tailored strategies to address distinct regional challenges. Non-EU countries must find ways to balance the losses in surveillance capabilities with the gains in compliance, while African nations are encouraged to build upon their momentum in prevention. In the EMR, there is an urgent need to focus on stabilizing healthcare systems to prepare for future crises. By recognizing these disparities and aligning regional strategies with global health priorities, stakeholders can work towards reinforcing the resilience and effectiveness of health security frameworks worldwide.^22,23

4.2 Interpreting regional disparities in health security rankings

The analysis of Global Health Security Index rankings among non-EU countries from 2019 to 2021 reveals a complex landscape of health security performance, with significant variations indicative of different responses to health security challenges and regional dynamics. As illustrated in Figure 3, thirteen countries (48.14%), including Belarus, Iceland, Kazakhstan, Liechtenstein, North Macedonia, San Marino, Tajikistan, Turkey, Uzbekistan, Azerbaijan, Georgia, Russia, and Montenegro, demonstrated improved rank values, signalling advancements in health security measures. These improvements underscore the effectiveness of targeted interventions and policy changes implemented in those nations, such as Montenegro’s successful health governance strategies and readiness to address public health threats.⁶²

In contrast, four countries (14.8%), including Andorra, Armenia, Norway, and the United Kingdom, maintained stable performance, indicating a consistent approach to health security that effectively mitigated the need for drastic policy shifts. However, a concerning 37% of countries, including Moldova, Israel, and the Kyrgyz Republic, saw declines in their rankings, which signals urgent areas for improvement in their health security frameworks. The declines in these countries are attributed to various factors, including the absence of comprehensive policies, inadequate biosecurity regulations, and insufficient accountability for international commitments.

Moldova’s decline, for instance, was characterized by a drop in rank due to the absence of policies concerning export/import restrictions and travel bans during infectious disease outbreaks, illustrating critical vulnerabilities in its health security preparedness.⁶³ Similarly, Israel faced significant reductions in foundational measures such as biosecurity regulations and immunization rates, leading to a decline in ranking. These gaps necessitate a re-evaluation of Israel’s health security infrastructure to enhance resilience and ensure effective responses to health crises.

Kyrgyzstan’s situation mirrors that of Moldova and Israel, as evidenced by similar declines in preparedness due to the absence of policies on travel restrictions and insufficient accountability for international commitments.⁶⁴ The 25-point decline in social stability further exacerbates the challenges faced by Kyrgyzstan in maintaining an effective health security framework. There is a critical need for Kyrgyzstan to prioritize health security improvements to better prepare for health emergencies and safeguard public health.

Collectively, these findings emphasize the need for a comprehensive understanding of the underlying causes contributing to the observed declines in health security rankings across non-EU countries. Targeted analyses are essential to develop effective strategies that can address these vulnerabilities, including enhancing regulatory frameworks, improving public health infrastructure, and fostering community engagement in health governance.⁶⁵ By acknowledging and addressing these critical areas, non-EU countries can bolster their health security frameworks, ultimately leading to stronger health outcomes and improved resilience against future public health threats.

4.3 Health security performance: Insights and patterns across five clusters

Using the final Entropy-CoCoSo assessment scores, non-EU countries were categorized into five distinct health security performance clusters during both individual sub-assessment years and the entire study period ( Table 4, Figure 4).

Table 4 and Figure 4 categorize countries into five clusters (High, Safe, Intermediate, Warning, and Dangerous) using K-means clustering of composite Entropy-CoCoSo scores. Cluster assignments reflect temporal stability or shifts in health security capabilities across the study period. These clusters, spanning from High to Dangerous, delineate varying levels of health security preparedness, offering a stratified perspective on systemic capabilities across non-EU countries.

To provide a clearer visualization of these country groupings, Figure 5 presents a radar chart of the clustering results.

4.3.1 Bridging gaps in Non-EU health security: Strategic interventions for cluster-specific risks

The analysis of health security indicators across five distinct clusters of non-EU countries reveals marked disparities and critical interdependencies that necessitate strategic interventions tailored to specific risks. Within the realm of prevention indicators, as shown in Figures 4, and 6, the high-performance cluster (Cluster 1), represented by countries like Armenia, Norway, and the United Kingdom, averaged a score of 87.48 in antimicrobial resistance (AMR) preparedness. This stands in stark contrast to Cluster 5 (Dangerous-Level), which includes nations such as Andorra, Monaco, San Marino, and Tajikistan, scoring significantly lower at 62.50. These disparities underscore the pressing need for targeted capacity-building initiatives in lower-performing regions.²³

Figure 4. Health security performance clustering results of Non-EU countries (Source: Authors).

This figure illustrates the K-Means Clustering results of non-EU European countries, encompassing the years 2019, 2021, and the entire period of 2017-2021. Countries are grouped into five clusters based on their health security profiles, with Cluster 1 representing exemplary performance and Cluster 5 representing critical performance. The figure also reveals patterns of similarity and dissimilarity among countries in terms of their health security capabilities.

Furthermore, the universal suboptimal readiness for zoonotic diseases across all clusters, particularly declining to a critical low of 14.53 in Cluster 5, highlights a collective vulnerability to emerging infectious diseases. The strikingly low biosafety and biosecurity metrics—21 and 0, respectively, in Cluster 5, as compared to scores of 63 and 67 in Cluster 1—point to systemic gaps that require urgent attention. The absence of dual-use research and responsible science practices outside of Cluster 1 further emphasizes the necessity for enhanced regulations and education throughout the entire region.

The evaluation of detection and reporting capabilities reveals a concerning downward trajectory of surveillance accessibility, laboratory systems, and epidemiological workforce capacity across clusters. The significant decline from an average score of 88.33 to 25 in surveillance accessibility reflects deteriorating capacities to monitor and respond to health threats effectively. Laboratory supply chains remain critically low, with Cluster 1 being the only group to achieve a score above 25. The extreme variance in epidemiological workforce metrics observed in Cluster 2 (1,148) highlights intra-cluster heterogeneity that requires more granular investigations to address the differences in capacity and deployment.

Rapid response capabilities exhibit stark disparities, with Cluster 1 leading at 52.80 while Clusters 4 and 5 lag behind with scores of 14.17 and 18.75, respectively. The universally inadequate testing of response plans indicates a significant gap in preparedness across all clusters. Notably, the integration of public health and security authorities is present only in Cluster 1 (100), emphasizing the crucial role of inter-agency collaboration in emergency preparedness.

The analysis of health system capacity reveals substantial gradients in health capacity and supply chain resilience, with scores ranging dramatically from 53.08 in some clusters down to 22.12 in others. The critical deficiencies in medical countermeasures—registered only at nominal levels in Clusters 1 and 2—point towards the need for focused investments in essential health resources and response mechanisms. Additionally, while infection control measures excel in Cluster 1, the moderate implementation seen in Clusters 4 and 5 indicates areas needing improvement.

Compliance metrics across clusters showed systemic shortcomings, particularly concerning joint external evaluations (JEE) and performance of veterinary services (PVS), which were universally neglected. Despite moderate cross-border collaboration in Clusters 1 to 4, Cluster 5’s scores indicate critical deficiencies that need immediate strategic attention. Analysis into compliance with International Health Regulations (IHR) further reflects the stark contrast between clusters, highlighting an urgent need for Cluster 5 to enhance its framework for adherence to international health standards.

The analysis of the risk environment unearthed paradoxical resilience patterns across clusters. While Cluster 5 exhibited high political-security stability (79.98) due to robust governance frameworks, it faced severe health system fragility. Conversely, Cluster 1 led in socioeconomic resilience (84.50), demonstrating the significant role of governance and infrastructure in shaping health security outcomes. The infrastructure adequacy in Cluster 5 (77.10), juxtaposed with its health system deficiencies, underscores the multifaceted nature of health security.⁶⁶

Those results stratified outcomes highlight the interdependence of governance, infrastructure, and specialized health capacities across clusters. The coexistence of robust infrastructure or political stability alongside health system fragility necessitates a nuanced approach to health security enhancement.^4,56 Prioritizing the integration of dual-use research, developing medical countermeasures, and fostering cross-sectorial coordination emerge as paramount for strengthening the overall resilience of health security frameworks in non-EU countries.⁶⁷ By addressing these targeted interventions, nations can better prepare for and respond to future public health challenges, ultimately contributing to a more secure and healthy global community.

4.3.2 Mapping Non-EU health security: Strengths, gaps, and policy implications across clusters

The mapping of health security performance across non-EU countries reveals significant strengths and gaps, with clear policy implications for each cluster. High-performing nations in Cluster 1, comprising Armenia, Norway, and the United Kingdom, illustrate the efficacy of integrated governance frameworks. Their strengths are evident in several key areas, including public health-security authority coordination, effective infection control measures, robust compliance with International Health Regulations (IHR), transparency in surveillance practices, high immunization coverage, and effective antimicrobial resistance (AMR) management. Additionally, these countries demonstrate socioeconomic resilience and solid international commitments. Serving as exemplars of best practices, they provide a replicable blueprint for global health security enhancement that other nations could adopt.²²

However, even among these top performers, persistent gaps exist, particularly in dual-use research governance, emergency response planning, laboratory supply chains, zoonotic disease mitigation efforts, case-based investigation protocols, medical countermeasure deployment, and healthcare worker communication during crises. There is also a concerning lack of participation in Joint External Evaluations (JEE) and Performance of Veterinary Services (PVS), which indicates systemic vulnerabilities that require targeted interventions to bolster overall preparedness.

In contrast, Cluster 2, which includes nations like Albania, Georgia, Iceland, Israel, Russia, Switzerland, and Turkey, showcases moderate capabilities. Unfortunately, only 3% of their indicators exceed high-performance thresholds (>80), compared to 24% in Cluster 1. The persistent critical deficiencies, identified in 34% of practices—such as emergency response operations, real-time surveillance capabilities, healthcare worker communication, and dual-use research oversight—underscore the urgent need for comprehensive, multidimensional assessments that identify and address latent risks within these countries.⁶⁷

Cluster 3, labelled as “Intermediate,” features countries like Belarus, Kazakhstan, and Kyrgyz Republic, which exhibit polarized performance metrics. Approximately 29.7% of their practices achieve “good” levels (60–80), yet 54% fall below 40. Strengths in immunization rates, reporting under the IHR, and access to healthcare infrastructure are significantly overshadowed by critical weaknesses in key areas, including case-based investigations, medical countermeasures, and cross-sectorial coordination. This landscape necessitates urgent structural reforms aimed at elevating health security standards and addressing the identified gaps.⁵⁴

Cluster 4, referred to as “Warning,” encompasses nations such as Azerbaijan, Bosnia and Herzegovina, and Turkmenistan, representing a group facing systemic collapse in 70% of health security indicators. Severe deficiencies are noted particularly in biosecurity, surveillance systems, emergency preparedness, and medical countermeasure effectiveness, despite isolated strengths in immunization coverage. The vulnerabilities within these nations highlight the need for immediate international support to avert catastrophic health system failures during public health crises.⁶⁸

Lastly, Cluster 5, designated as “Dangerous,” includes Andorra, Monaco, San Marino, and Tajikistan. This group presents a paradox where competitive trade and travel restriction policies (e.g., scoring 90.63) coexist with critical failures across 70% of health security domains. Their disproportionately high political-security stability scores (79.98) obscure the underlying fragility of their health systems, signalling a pressing need for comprehensive governance overhauls. Aligning infrastructure investments with public health priorities is imperative to build a more resilient health security framework.⁶⁹

In addition, the disparities uncovered through this mapping exercise underscore the urgent need for targeted interventions across clusters to fortify health security. High-performing nations should continue to enhance their existing frameworks while sharing best practices to assist others. Meanwhile, lower-performing countries require international collaboration, increased funding, and capacity building to address specific weaknesses that impede their ability to effectively manage public health threats. By prioritizing these strategic interventions, non-EU nations can progress toward a more unified and effective global health security architecture.

Specific interventions are recommended for low-performing clusters to transform these findings into actionable policies. For countries in Cluster 4 (“Warning”), which are on the brink of systemic collapse, immediate international support is essential. This support should prioritize foundational capabilities, such as the implementation of basic biosecurity protocols, the supply of essential laboratory equipment, and the training of frontline healthcare workers in disease surveillance. For nations in Cluster 5 (“Dangerous”), which display a paradox of high political stability but fragile health systems, interventions should concentrate on governance reform. This involves establishing independent public health agencies, securing budgets for health security, and creating clear accountability mechanisms to ensure that infrastructure investments align with public health needs. Institutionalizing cross-cluster mentorship, where nations from Clusters 1 and 2 offer technical assistance to those in Clusters 4 and 5, should be pursued to facilitate knowledge transfer and build capacity.