A Study on Key Factors Affecting the Resilience of Emergency Logistics Supply Chains: A Hybrid Fuzzy DEMATEL-ISM-MICMAC Approach

Abstract

Against the backdrop of global climate change, frequent public health crises, and escalating geopolitical conflicts, the stable operation of emergency logistics supply chains faces severe challenges. Building a resilient system that combines disturbance resistance and adaptability has become an urgent necessity. This paper, grounded in the evolution of resilience theory, clearly defines the meaning of emergency logistics supply chain resilience. It systematically identifies and constructs an indicator system comprising 17 influencing factors across four dimensions: Resistance, Responsiveness, Adaptability, and Development Capacity. Employing a hybrid fuzzy DEMATEL-ISM-MICMAC approach, the study quantifies causal relationships and hierarchical structures among factors while analyzing their driving forces and dependency attributes. Findings reveal that infrastructure development, emergency plan integrity, talent cultivation, financial safeguards, and regulatory support constitute core critical factors influencing emergency logistics supply chain resilience. Among these, regulatory support and financial safeguards form the fundamental pillars underpinning the system’s operation. The multidimensional influence factor framework and hybrid analytical method developed in this study not only enrich the theoretical research system on emergency logistics supply chain resilience but also provide scientific decision-making references and practical guidance for policymakers and industry practitioners to formulate targeted resilience enhancement strategies.

1. Introduction

Since the beginning of the 21st century, frequent global climate change, geopolitical conflicts, and the normalization of public health crises have posed significant threats to human society. According to World Bank data, between 2014 and 2024, major global emergencies led to frequent supply chain disruptions, resulting in economic losses that exceeded 3.5 times the GDP growth rate during the same period. Examples include the global shortage of electronic components following the Fukushima nuclear incident in Japan, the disordered cross-border allocation of medical supplies during the early stages of the COVID-19 pandemic, and the disruptions in energy and food supply chains caused by the Russia–Ukraine conflict. These events have resulted in substantial casualties and economic losses, while also presenting considerable challenges to the global emergency supply system. China is among the countries most frequently affected by severe natural disasters [1]. Due to its climate and geography, China experiences frequent geological and meteorological hazards, including earthquakes, floods, and typhoons [2]. These hazards pose significant risks to the lives and property of the Chinese population and to national development. Such circumstances highlight the structural weaknesses and vulnerabilities of traditional supply chain systems when faced with extreme uncertainty. Previous disaster relief operations have demonstrated that improving the stability and reliability of emergency logistics supply chains is essential for minimizing disaster losses and ensuring public safety. As a result, a growing number of practitioners and scholars advocate for the development of resilient emergency logistics supply chain systems [3,4,5].

Recognizing the critical importance of this issue, countries and international organizations worldwide have placed enhancing supply chain resilience at the core of their policy agendas. The United Nations Sendai Framework for Disaster Risk Reduction 2015–2030 explicitly emphasizes strengthening pre-disaster preparedness and building disaster resilience; international institutions such as the World Bank have begun vigorously promoting the development of resilient supply chains. The Chinese government attaches even greater importance to this matter. The Report to the 20th CPC National Congress explicitly calls for “strengthening the resilience and security of industrial and supply chains” and “enhancing urban infrastructure to build resilient smart cities,” stressing that this is an essential requirement for establishing a new development paradigm, a necessary step toward achieving high-quality development, and a vital component of national security. The National Emergency System Plan for the 14th Five-Year Plan explicitly calls for “strengthening the emergency material supply system,” making clear arrangements for building smart emergency response capabilities and enhancing supply chain coordination and resilience. This provides strong policy impetus for related research and practice.

However, current research on emergency logistics supply chain resilience has significant limitations. There is a lack of a systematic framework for identifying the factors influencing emergency logistics supply chain resilience. Second, there is insufficient in-depth analysis of the intrinsic interaction pathways and hierarchical relationships among various influencing factors. Therefore, the core issue addressed by this study is: identifying the key factors and core drivers influencing the resilience of emergency logistics supply chains in uncertain and high-risk environments and constructing a theoretical model of resilience-influencing factors. A hybrid fuzzy DEMATEL-ISM-MICMAC approach is employed to analyze the causal interaction mechanisms, hierarchical relationships, and driving-dependency attributes among these factors. Based on the driving-dependent attributes of these factors, guide the formulation of corresponding strategies to achieve a shift from passive response to proactive evolution.

2. Literature Review

2.1. Resilience of Emergency Logistics Supply Chains

Emergency logistics supply chains refer to the planning, implementation, and control processes for materials, funds, and information from supply points to demand destinations, aimed at achieving humanitarian relief in response to sudden public emergencies [6]. Early research viewed emergency logistics supply chains as an application of commercial supply chains in special contexts, but subsequent scholars recognized fundamental differences in their objectives. While commercial supply chains pursue cost minimization or profit maximization, the core objective of emergency logistics supply chains is humanitarianism—minimizing loss of life and human suffering. This leads to fundamentally different performance measurement criteria [7].

This divergence in objectives results in several distinctive features of emergency logistics supply chains. Demand increases rapidly within very short periods after catastrophic events, accompanied by significant uncertainty regarding the types and quantities of goods required [8]. Emergency supplies are frequently sourced through donations, which often leads to a substantial mismatch between supply and demand. This mismatch can cause a “secondary disaster,” where surplus unwanted goods accumulate while essential items remain in short supply [9]. The “last-mile” delivery process is particularly complex due to damaged infrastructure, disrupted transportation, and security risks, frequently necessitating the use of informal networks or military forces to ensure delivery [10].

Early research focused on operational aspects such as designing optimal emergency logistics networks [11], siting material reserve warehouses [12], and planning transportation routes [13]. Analysis of emergency logistics supply chain effectiveness also became a significant research direction. Wu et al. [14] proposed a three-dimensional evaluation framework covering effectiveness, efficiency, and equity. This work laid the foundation for later studies. As research progressed, scholars examined collaborative management issues within emergency logistics supply chains. The main challenges identified included involving diverse stakeholders with conflicting interests, maintaining information silos that impede data sharing, and frequent coordination failures among responsible entities. Wankmüller et al. [15] investigated establishing effective cross-departmental collaboration mechanisms and information-sharing platforms. In recent years, research has shifted toward digital technologies [16] and sustainable development [17] within emergency logistics supply chains.

The term “resilience” originated in physics, describing a material’s ability to absorb energy and deform without fracturing under external forces. Canadian ecologist Holling [18] first used it to discuss ecosystem stability. In 2004, Martin et al. [5] applied resilience to supply chain research, defining it as “the ability of a supply chain network to anticipate, prepare for, respond to, and recover from disruptions to its original or improved operational state.” Wieland et al. [19] later described supply chain resilience as the capacity to maintain, adapt, or transform in response to change, incorporating both engineering and socio-ecological perspectives.

Research on supply chain resilience has evolved from simple conceptual definitions to in-depth exploration of its stages, dimensions, and mechanisms in recent years, driven by increased risks of supply chain disruptions stemming from global political, economic, and social issues. Brandon et al. [20] proposed that visibility and collaboration are core factors influencing supply chain resilience; Pavlov et al. [21] suggested that supply chain resilience comprises four dimensions: robustness, agility, redundancy, and resource reconfiguration capability. Blackhurst et al. [22] emphasized that achieving supply chain resilience requires balancing control and vulnerability, with flexibility and redundancy as key enablers. Kamalahmadi et al. [23] confirmed that moderate redundancy and flexible structures significantly enhance system robustness and resilience, advocating for trade-offs between redundancy costs and resilience benefits. Building upon this, Faruquee et al. [24] defined distinct phases of supply chain resilience: passive resilience (reactive recovery) and active resilience (proactive preparedness). They contend that highly resilient organizations excel not only in responding and recovering but also in anticipating risks and institutionalizing knowledge gained from disruptions, thereby creating a virtuous cycle of “learning–adapting–evolving”.

Jabbour et al. [25] define emergency logistics supply chain resilience as the capability of an emergency logistics supply chain system to maintain core functions without interruption, achieve rapid recovery, and learn and evolve in response to extreme disturbances. In recent years, the surge in natural disasters and international conflicts has heightened risks and uncertainties, making the external environment for emergency logistics supply chains increasingly complex and severe. Consequently, advancing the development of routine security assurance mechanisms and emergency support systems to enhance the resilience of emergency logistics supply chains has become critically important [26].

Ge et al. [27] proposed that increasing the number of transportation links, enhancing information monitoring and early warning capabilities, and improving contingency plan precision significantly boost the resilience of emergency logistics supply chains. Wang et al. [28] argued that governments play a central role in emergency logistics supply chains, making the completeness of laws and regulations, cross-administrative coordination mechanisms, and funding guarantee systems key prerequisite factors. Nunes et al. [29] proposed that emergency logistics supply chains are essentially temporary, dynamic “response networks,” whose resilience depends not only on individual organizational capabilities but also on the overall coordination efficiency of the network. Building upon this, Wang et al. [30] propose that effectively integrating social forces—including enterprises, non-governmental organizations, and volunteers—to form a “government-market-society” collaborative emergency network is a crucial pathway to enhancing emergency logistics supply chain resilience. It is worth emphasizing that emergency environments are highly uncertain, and fixed contingency plans often prove ineffective [31]. Therefore, a highly resilient emergency logistics supply chain system must possess robust organizational dynamic capabilities—namely, the ability to sense environmental changes, seize opportunities, and reconfigure resources [32]. Concurrently, Zaoui et al. [33] contend that a highly resilient emergency logistics supply chain system must inherently be a learning system, capable of continuously absorbing lessons from internal and external shocks and transforming them into structural optimization and functional upgrades.

Despite significant progress in existing research on supply chain resilience, substantial research gaps remain when applying these theories to emergency logistics scenarios. First, existing research predominantly focuses on the trade-off between supply chain efficiency and cost, with resilience definitions primarily confined to passive aspects like disturbance resistance and recovery to original states. However, emergency logistics supply chains confront extreme uncertainty risks, and their humanitarian objectives demand systems not only recover but also achieve developmental upgrades through learning during disasters. Current metric systems generally lack consideration for this developmental capacity and dynamic adaptability. Second, while existing research has identified some influencing factors, these are often treated as isolated variables or simple linear relationships. This approach risks producing superficial strategic recommendations that fail to address the core of systemic transformation. Emergency logistics systems exhibit high complexity and coupling, with resilience emerging from the interplay between hard infrastructure and soft management capabilities. Consequently, further research is needed to establish an appropriate framework.

2.2. Analysis Methods for Influencing Factors

Influencing factor analysis methods, including the Analytic Hierarchy Process (AHP) [34], the Fuzzy Technique for Order Preference by Similarity to Ideal Solution (FTOPSIS) [35], Structural Equation Modeling (SEM) [36], combined weighting methods based on game theory [37], and hybrid fuzzy-probabilistic approaches [21], have been extensively utilized in diverse disciplines. Despite their widespread application, these methods frequently lack the capacity to conduct detailed attribute analysis of individual indicators and often do not sufficiently evaluate the intensity, significance, or hierarchical structure of relationships among factors. Supply chain systems, in particular, demonstrate nonlinear coupling characteristics, and their resilience is inherently multi-dimensional and structurally complex [38]. Consequently, to accurately identify the key variables affecting resilience in emergency logistics supply chains and to map the influence pathways among these variables, it is necessary to develop approaches that address both the identification of critical variables and the elucidation of their interrelationships.

In exploring evaluation methods for influencing factors, scholars have conducted extensive research. Among these, the DEMATEL-ISM method has gained widespread application. Guo et al. [39] utilized the DEMATEL-ISM method to analyze pathways for mitigating supply chain vulnerability under uncertain conditions. Feng et al. [40] employed a DEMATEL-ISM combination approach to identify factors influencing the resilience of prefabricated supply chains and proposed pathways to enhance resilience. However, the traditional DEMATEL-ISM method also has limitations, exhibiting strong subjectivity and lacking the ability to quantify fuzzy linguistic descriptions. Therefore, replacing the traditional DEMATEL method with the fuzzy DEMATEL method—which converts expert semantics into corresponding triangular fuzzy numbers—can address these shortcomings. This approach enhances the objectivity of results, avoids reliance on discrete scores inherent in traditional methods, and scientifically determines the importance of factors. Its validity has been demonstrated in numerous studies. Wang et al. [41] analyzed the influencing factors and their mechanisms on coal-fired power supply chain resilience using the fuzzy DEMATEL-ISM approach. Liu et al. [42] employed an integrated fuzzy DEMATEL-ISM method to examine causal relationships and hierarchical logic among factors affecting cross-border e-commerce supply chain resilience. Furthermore, the MICMAC analysis method better reflects the interdependencies and driving forces among various influencing factors [43]. Therefore, combining the aforementioned methods facilitates more accurate problem analysis, and hybrid approaches have been applied in diverse studies. Feng et al. [44] employed the fuzzy DEMATEL-ISM-MICMAC method to analyze factors influencing employee engagement in green behaviors and identified key determinants. Liu et al. [45] employed the fuzzy DEMATEL-ISM-MICMAC method to analyze the influencing factors of major transportation infrastructure projects and proposed pathways and methods for enhancing resilience.

In summary, although the hybrid fuzzy DEMATEL-ISM-MICMAC method has been applied in supply chain research across manufacturing, construction, and other sectors, its introduction into the field of emergency logistics supply chain resilience remains significantly necessary and innovative. Unlike typical commercial supply chains, emergency logistics supply chains involve multiple stakeholders, including governments, military forces, enterprises, and non-governmental organizations, with influencing factors exhibiting stronger fuzziness and nonlinear characteristics. Existing single-method approaches (e.g., TOPSIS or AHP) primarily focus on static weight calculations, struggling to dynamically reveal the complex causal-driven structures among factors. Current quantitative research (e.g., SEM) also fails to uncover these deep, nonlinear driving structures. Furthermore, traditional regression analysis is constrained by the scarcity of extreme disaster data, making large-sample empirical studies challenging. Concurrently, single-structured methods exhibit inherent limitations. Traditional ISM methods overlook subtle variations in the degree of influence between factors. When handling a large number of factors, the complex causal network diagrams generated by traditional DEMATEL methods often become overly chaotic.

To address the aforementioned research gap and systemic complexity, this paper aims to construct an evaluation index system for emergency logistics supply chain resilience and proposes a hybrid fuzzy DEMATEL-ISM-MICMAC analytical framework. This study does not merely transplant the method but, grounded in the unique “government-led, multi-stakeholder collaborative” nature of emergency logistics, delves into a core question: how to identify the fundamental drivers that influence the entire emergency system’s transition from passive disaster response to proactive evolution, and to decipher the pathway mechanisms through which these drivers translate into system resilience via intermediate variables. This hybrid approach provides a systematic, dynamic solution by quantifying uncertainty, deconstructing hierarchical relationships, and dynamically identifying sensitivities. This deep structural analysis tailored to specific contexts constitutes the unique empirical insight distinguishing this study from existing literature, while also offering scientific theoretical support and decision-making references for formulating targeted resilience enhancement strategies.

3. Materials and Methods

3.1. Resilience Framework for Emergency Logistics Supply Chains

The driving mechanisms and optimization pathways influencing emergency logistics supply chain resilience are an emerging research area. This area remains insufficiently explored. Drawing on supply chain resilience theory and the characteristics of emergency logistics supply chains, this study defines emergency logistics supply chain resilience. It is the capability of a supply chain system to stay vigilant, absorb disruptions quickly, and adjust responses during emergencies to keep logistics stable. It also involves prompt resumption of normal operations after an emergency and enhancing system functionality by summarizing and optimizing the response process.

Resilience theory emphasizes comprehensive management of risks and emergencies throughout the entire process—before, during, and after an incident—aiming not only to restore systems to their original state but also to elevate them to a new, more adaptive, stable state. By reviewing existing literature, the authors found that academic research on supply chain resilience dimensions primarily focuses on core capabilities such as preparedness [46,47], responsiveness [46,48], and recovery [49,50], primarily focusing on a system’s ability to withstand, absorb, and rapidly recover from known or sudden disruptions. These studies provide important theoretical foundations for this paper, but their perspectives tend to focus on maintaining stability and restoring the original state before and after disturbances occur, which can be categorized as static resilience or passive resilience.

However, emergency logistics supply chains demand heightened resilience due to their unique humanitarian core objectives and extremely uncertain external environments. On one hand, the complex and unpredictable disaster scenarios they confront require systems to possess forward-looking adaptability—the ability to proactively adjust their structures, strategies, and processes to address unknown, continuously evolving challenges [51]. On the other hand, emergency operations often entail high human and societal costs, where mere “restoration to original state” may mean repeated exposure to identical risks. Therefore, the system must possess development capacity—the ability to systematically learn from past disruptions, accumulate knowledge, and translate experience into structural optimization and capability upgrades, achieving a virtuous cycle of “post-event learning and pre-event evolution” [52,53,54]. This developmental capacity stems from continuous post-crisis reviews of emergency processes and outcomes, the establishment of knowledge management systems, and the institutionalization of learning outcomes within organizational processes, technical standards, and collaborative networks. These two capabilities together form the core of dynamic resilience, enabling systems not only to withstand immediate shocks but also to achieve incremental improvements in resilience over the long term.

To strengthen the theoretical basis for dimensional classification, this paper draws upon the “perception-capture-reconstruction” framework from dynamic capability theory and the “single-loop learning-double-loop learning” mechanism from organizational learning theory to further distinguish between adaptive capability and developmental capability: Adaptive capacity focuses more on flexible adjustments to processes and strategies within existing frameworks and objectives, corresponding to single-loop learning and capability capture. Developmental capacity emphasizes reflection and transformation of existing objectives, mental models, and system structures, corresponding to double-loop learning and capability reconstruction. In emergency logistics supply chains, adaptive capacity manifests as real-time adjustments to resource allocation paths and collaboration models during disaster response to address unforeseen situations; development capability manifests as post-disaster system reviews, knowledge institutionalization, process reengineering, and collaborative network optimization to elevate the system’s long-term resilience baseline. Thus, developmental capability is not a subset of adaptive capability or general learning behavior but represents a higher-order, more fundamental process of organizational transformation and strategic renewal. Its core lies in translating learning outcomes into systematic, structured capability evolution.

Therefore, to more comprehensively and deeply characterize the integrated capability of emergency logistics supply chains to achieve humanitarian objectives in highly uncertain environments, this study expands and integrates upon classical dimensions to construct a four-dimensional analytical framework encompassing “Resistance, Responsiveness, Adaptability, and Development Capacity,” as illustrated in Figure 1. This framework not only encompasses the core elements of traditional resilience theory in addressing known shocks but also emphasizes the system’s proactive adaptive capacity in the face of uncertainty and the fundamental drivers for achieving long-term resilience enhancement. The four-dimensional framework aims to systematically deconstruct the complex nature of emergency logistics supply chain resilience throughout the entire process—from passive defense to active adaptation to evolutionary learning—thereby better aligning with its unique requirements for coping with extreme and uncertain environments. This provides a solid theoretical foundation for the present study on emergency logistics supply chain resilience.

3.2. Identification of Influencing Factors

The keywords “supply chain resilience,” “emergency logistics supply chain,” and “influencing factors” were used to screen literature in databases including Web of Science, Google Scholar, and CNKI, covering the period from 2010 to 2025. The initial search identified 337 relevant articles. Following a review of titles, abstracts, and research domains, 24 articles were deemed valid for inclusion. Research on the resilience of emergency logistics supply chains remains limited. Existing studies on supply chain resilience [55,56,57,58,59,60,61,62,63,64,65] were referenced to identify influencing factors. These factors and their annotations are summarized in

Table 1. Factors Influencing the Resilience of Emergency Logistics Supply Chains and Annotations.

Dimension	Influencing Factors	Commentary	Reference
Resistance	S1 Risk forecasting and prediction	Identify, analyze, and predict risks, including potential natural disasters and public health incidents.	[55,61,62]
	S2 Infrastructure construction	Redundancy and reliability design for transportation networks, communication facilities, warehousing centers, and other infrastructure.	[59,60]
	S3 Equipment maintenance	Daily inspection, upgrade, and maintenance of equipment such as warehousing facilities, transportation vehicles, and communication devices.	[55,64]
	S4 Natural environment compatibility	Whether the site selection, design, and construction standards for key nodes in the supply chain are compatible with the local natural environment.	[55]
Responsiveness	S5 Completeness of the emergency plan	In the process of formulating and implementing emergency response plans, all possible scenarios and risks were thoroughly considered, and corresponding countermeasures were proposed.	[55,58,61]
	S6 Decision making and responsiveness	Rapidly conduct disaster assessments, formulate recovery strategies, and issue scientifically sound decisions.	[55,58,60,61]
	S7 Transportation capacity	The rapid dispatch capability of transportation vehicles, the integrity of transported goods, and the punctuality of deliveries.	[62]
	S8 Logistical maintenance and storage management	The efficiency of logistics support and the management of receiving, storing, sorting, and distributing relief supplies.	[55,58]
	S9 Social stability	The soundness of social order, public services, and community governance provides a secure macro environment for the operation of emergency logistics supply chains.	[58,60]
Adaptability	S10 Data sharing and coordination	Real-time data exchange and business collaboration among all supply chain participants to prevent information delays.	[56,57,59,61]
	S11 Dynamic adjustment capability	The supply chain possesses alternative options and flexibility, enabling the ability to swiftly switch suppliers, alter distribution models, and select substitute materials.	[56,57,58]
	S12 Simulation and drill	Through drills, cultivate experience and team cohesion to rapidly adapt and respond to emergencies.	[64]
	S13 Integration of emerging technologies	Integrate cutting-edge technologies such as AI, the IoT, blockchain, and big data into emergency logistics supply chains to adapt to complex new environments.	[59,61]
Development Capacity	S14 Learning and innovating	Establish a systematic post-event review mechanism to draw lessons from past practices and drive supply chain adjustments and optimization.	[57,60,61]
	S15 Talent cultivation	Develop the professionalism and emergency response capabilities of personnel involved in supply chain design, production, transportation, and management.	[59,63]
	S16 Legal and regulatory support	Are the laws, regulations, and policies established by the state for emergency situations sound and clear.	[56,65]
	S17 Funding guarantee	The adequacy of financial arrangements such as contingency fund reserves, emergency funding mechanisms, and R&D investment.	[58,60]

.

To ensure the reliability of the indicator system for factors influencing the resilience of emergency logistics supply chains, this study employed expert interviews and survey methods to refine the influencing factors. Ultimately, 17 factors were identified, and an indicator system for emergency logistics supply chain resilience was constructed based on four dimensions: Resistance, Responsiveness, Adaptability, and Development Capacity (Figure 2).

3.3. Methodology

Gabus et al. [66] initially introduced the Decision Experiment and Evaluation Laboratory (DEMATEL) method, which utilizes graph theory and matrix tools for systematic factor analysis. The primary objective of DEMATEL is to quantify the mutual influence among factors within a system and to visualize complex relationships using causal diagrams, thereby elucidating the system’s underlying logic [67]. Fuzzy set theory addresses ambiguous relationships among elements by modeling the human brain’s approach to processing imprecise information [68]. In this study, fuzzy set theory is integrated with the DEMATEL method by incorporating triangular fuzzy numbers into the traditional DEMATEL framework. This approach translates expert judgments into equivalent triangular fuzzy integers, which reduces the subjectivity associated with expert scoring.

The Structural Interpretation Model (ISM), proposed by Warfield [69], is a widely utilized tool in systems engineering analysis. This model employs directed graphs and matrix-based methods to decompose complex systems into multiple constituent factors. By examining the logical relationships among driving factors, ISM constructs a multi-level hierarchical model characterized by clear structure and distinct layers. The Matrix of Inter-Causal Multiplication (MICMAC) is a complementary method for analyzing the mutual influences among factors within complex systems. MICMAC quantifies the strength of indirect effects between factors through matrix multiplication, classifies variables, and identifies key drivers. Its primary function is to use iterative calculations of the reachability matrix to quantify dependencies and driving forces among system elements.

This paper constructs a hybrid fuzzy DEMATEL-ISM-MICMAC method that integrates the strengths of all three approaches: fuzzy DEMATEL enhances the objectivity and reliability of measuring influence relationships among factors, ISM further parses the hierarchical structure of the system, and MICMAC supplements the analysis of factors’ driver-dependent attributes. This integration enables a comprehensive, multidimensional analysis of a system’s factors—covering their importance, hierarchical structure, and attribute characteristics. It overcomes the limitations of single methods when handling complex systems, enabling analysis that spans from micro-level causal relationships to macro-level system architecture. This significantly enhances the systematicity and reliability of analytical outcomes.

3.4. Construction of a Fuzzy DEMATEL-ISM-MICMAC Integrated Model

This study employs a hybrid fuzzy DEMATEL-ISM-MICMAC method to analyze the relationships among factors influencing the resilience of emergency logistics supply chains. Figure 3 illustrates the methodological logic flow of this research, designed as a closed-loop system analysis framework. This methodology comprises three tightly coupled modules forming a rigorous logical chain:

Module One serves as the input layer, employing fuzzy set theory to convert qualitative expert knowledge into quantitative triangular fuzzy numbers. This addresses the challenge of “information imprecision” in emergency scenarios, generating a standardized direct influence matrix. Module Two functions as the core processing layer, utilizing DEMATEL to compute the comprehensive influence between factors and quantify causal strength. Subsequently, by setting a statistical threshold λ, the continuous influence matrix is transformed into the discrete adjacency matrix required for ISM. Module Three serves as the output and validation layer. ISM outputs a multi-level hierarchical structure model, revealing the system’s vertical hierarchy; MICMAC outputs a driver-dependent quadrant diagram, revealing the horizontal attributes of factors. The two mutually corroborate, ensuring that the ultimately identified key factors possess both causal root causes and high systemic driving forces.

The methodological model proposed in this study is clear and transparent, enhancing the rigor and reproducibility of the research.

Step 1: Identify the set of influencing factors $\mathrm{S}=\{{\mathrm{S}}_1,{\mathrm{S}}_2,\dots ,{\mathrm{S}}_{\mathrm{n}}\}$. Obtain the direct influence matrix $\mathrm{P}=[{\mathrm{p}}_{\mathrm{i}\mathrm{j}}{]}_{\mathrm{n}\times \mathrm{n}}$ through questionnaires and expert interviews, where ${\mathrm{p}}_{\mathrm{i}\mathrm{j}}$ represents the direct influence of factor i on factor j. When i = j, ${\mathrm{p}}_{\mathrm{i}\mathrm{j}}=0$.

Step 2: Construct the expert linguistic evaluation set and define the linguistic variables used by the expert panel, as shown in

Table 2. Semantic Conversion Table.

Language Variable	Expert Scoring	Triangular Fuzzy Number
No impact	0	(0, 0, 0.25)
Weak impact	1	(0, 0.25, 0.50)
Moderate impact	2	(0.25, 0.50, 0.75)
Strong impact	3	(0.50, 0.75, 1.00)
Extremely strong impact	4	(0.75, 1.00, 1.00)

. Based on the linguistic variables established by Wang et al. [70] for the expert panel, convert the initial direct influence matrix into corresponding triangular fuzzy numbers. The triangular fuzzy number X can be represented as $(l,\mathrm{m},\mathrm{r})$. Let ${\mathrm{X}}_{\mathrm{i}\mathrm{j}}^{\mathrm{k}}=(l_{ij}^k,m_{ij}^k,r_{ij}^k)$ denote the triangular fuzzy evaluation of factor i relative to factor j provided by expert k, where $l$ is the conservative value (lower bound of the fuzzy number), m is the estimated value, and r is the optimistic value (upper bound of the fuzzy number), with $l⩽\mathrm{m}⩽\mathrm{r}$.

Step 3: Apply the CFCS method for defuzzification.

Traditional centroid methods are prone to losing the left-right distribution information of fuzzy numbers during defuzzification. The CFCS method employs standardization based on fuzzy maximum and minimum values to more sensitively capture the upper and lower boundary differences within the “fuzzy interval” of expert evaluations. This enables it to more accurately reflect experts’ true preference tendencies when facing uncertainty, ensuring the precision of input data. Therefore, this study selects the CFCS method for defuzzification processing.

The specific steps are as follows:

(1) Standardize the triangular fuzzy numbers.

$$\left\{\begin{array}{c}{a}_{ij}^k=[l_{ij}^k-min(l_{ij}^k\left)\right]/{\mathsf{\Delta }}_{min}^{max}\hfill \\ b_{ij}^k=[m_{ij}^k-min(m_{ij}^k\left)\right]/{\mathsf{\Delta }}_{min}^{max}\hfill \\ c_{ij}^k=[r_{ij}^k-min(r_{ij}^k\left)\right]/{\mathsf{\Delta }}_{min}^{max}\hfill \\ {\mathsf{\Delta }}_{min}^{max}=max\left(r_{ij}^k\right)-min\left(l_{ij}^k\right)\hfill \end{array}\right.$$

(1)

In the formula, $a_{ij}^k$, $b_{ij}^k$ and $c_{ij}^k$ represent the normalized left-hand value, middle value, and right-hand value, respectively; ${\mathsf{\Delta }}_{min}^{max}$ denotes the difference between the right-hand value and the left-hand value.

(2) Standardize the left-hand side and right-hand side values.

$$\left\{\begin{array}{c}{u}_{ij}^k=b_{ij}^k/(1+b_{ij}^k-a_{ij}^k)\\ v_{ij}^k=c_{ij}^k/(1+c_{ij}^k-b_{ij}^k)\end{array}\right.$$

(2)

In the formula, $u_{ij}^k$ and $v_{ij}^k$ represent the normalized values of the left-hand side and right-hand side, respectively.

(3) Calculate the normalized total value.

$$w_{ij}^k={\displaystyle \frac{u_{ij}^k\cdot (1-u_{ij}^k)+(v_{ij}^k{)}^2}{1-u_{ij}^k+v_{ij}^k}}$$

(3)

(4) Calculate the precise value $d_{ij}^k$ of the triangular fuzzy judgment given by expert k.

$$d_{ij}^k=min\left(l_{ij}^k\right)+w_{ij}^k\cdot {\mathsf{\Delta }}_{min}^{max}$$

(4)

Calculate the standard precision value $d_{ij}$ of the pth expert’s evaluation to determine the fuzzy direct influence matrix D.

$$d_{ij}={\displaystyle \frac{1}{p}}\cdot \sum _{k=1}^p{d}_{ij}^k$$

(5)

$$D=\left[\begin{array}{cccc}0& d_{12}& \dots & d_{1n}\\ d_{21}& 0& \dots & d_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ d_{n1}& d_{n2}& \dots & 0\end{array}\right]$$

(6)

Step 4: Calculate the normalized direct influence matrix G

Perform normalization on the fuzzy direct influence matrix to obtain the normalized direct influence matrix G.

$$G={\displaystyle \frac{D}{max\left(\sum _{j=1}^n·d_{ij}\right)}}$$

(7)

This normalization process ensures that each element in matrix G falls within the range [0, 1].

Step 5: Calculate the Comprehensive Influence Matrix T

By summing the direct and indirect influences between factors, the comprehensive influence of each factor relative to the highest-ranked factor within the system is determined, yielding:

$$T=(G+G^2+\dots +G^n)=G{\displaystyle \frac{1-G^{n-1}}{1-G}}=G(1-G{)}^{-1}$$

(8)

Step 6: Calculate the influence degree, affected degree, centrality, and causality degree for each factor

The influence degree $D_i$ refers to the sum of all elements in each row of matrix T, reflecting the comprehensive influence of the corresponding factor on all other factors in the system. The affected degree $E_i$ denotes the sum of all elements in each column of matrix T, indicating the comprehensive influence of that factor on all other factors.

Centrality $M_i$ reflects a factor’s relative position within the evaluation system and the magnitude of its role. Causality $R_i$ reflects the logical relationships among factors. When causality is greater than 0, it indicates that the factor exerts a strong influence on others and is termed a causal factor. Conversely, if causality is less than 0, the factor is a resultant factor.

$$\left\{\begin{array}{c}{D}_i={\displaystyle \sum _{j=1}^n}{t}_{ij},i=1,2,\dots n\hfill \\ E_i={\displaystyle \sum _{i=1}^n}{t}_{ij},j=1,2,\dots n\hfill \\ M_i=D_i+E_i,i=1,2,\dots n\hfill \\ R_i=D_i-E_i,i=1,2,\dots n\hfill \end{array}\right.$$

(9)

Step 7: Plot the Cause-Effect Diagram

This study plots the cause-effect diagram with centrality $M_i$ as the horizontal axis and responsibility $R_i$ as the vertical axis.

Step 8: Calculate the Overall Influence Matrix $\mathrm{H}=({\mathrm{h}}_{\mathrm{i}\mathrm{j}}{)}_{\mathrm{n}\times \mathrm{n}}$

$$\mathrm{H}=\mathrm{T}+\mathrm{E}$$

(10)

Step 9: Determine the threshold λ, compute the adjacency matrix F and the reachability matrix K

A threshold λ is introduced to eliminate relationships between factors with minor influence, thereby clarifying the hierarchical structure of the system. The value of λ ranges between [0, 1]. The threshold setting directly impacts the interpretability of the system structure. An excessively low threshold may result in overly complex relationships, making it difficult to identify key pathways; conversely, an excessively high threshold may overlook significant associations. To avoid biases arising from subjective threshold determination, this study employs the statistically grounded “mean plus standard deviation method” to determine the λ value. By applying thresholding to the overall influence matrix H, the adjacency matrix F is obtained.

$$f_{ij}=\left\{\begin{array}{c}1 h_{ij}\ge \lambda (i,j=\mathrm{1,2},\dots ,n)\\ 0 h_{ij}<\lambda (i,j=\mathrm{1,2},\dots ,n)\end{array}\right.$$

(11)

To convert to the reachability matrix required for the ISM method, the adjacency matrix F is added to the identity matrix to obtain the multiplication matrix B. Based on the arithmetic properties of Boolean matrices, several Boolean operations are performed on this matrix, yielding the reachability matrix K.

$$K={\left(F+E\right)}^{n+1}={\left(F+E\right)}^n\ne {\left(F+E\right)}^{n-1}\ne F+E$$

(12)

Step 10: Construct a multi-level hierarchical structure model

$$\left\{\begin{array}{c}R\left(S_i\right)=\left\{S_j\right|s_j\in S,k_{ij}=1\}\hfill \\ A\left(S_i\right)=\left\{S_j\right|s_j\in S,k_{ji}=1\}\hfill \\ C\left(S_i\right)=Z\left(S_i\right)\cap A\left(S_i\right)\hfill \end{array}\right.$$

(13)

Determine the reachable set $Z\left(S_i\right)$, the antecedent set $A\left(S_i\right)$ and the common set $C\left(S_i\right)$. Repeat Step 10 until all factors are hierarchically partitioned.

Step 11: MICMAC Analysis

Calculate the driving force $Q_i$ and dependency $Y_i$. Plot the MICMAC results analysis diagram.

$$\begin{array}{c}{ Q}_i={\displaystyle \sum _{i=1}^{n+1}}{k}_{ij}\\ { Y}_i={\displaystyle \sum _{j=1}^{n+1}}{k}_{ij}\end{array}$$

(14)

4. Results

4.1. Data Collection

This study employs a five-point scale ranging from 0 to 4 to quantify the degree of association among various influencing factors. A score of 0 indicates no influence, 1 denotes weak influence, 2 represents moderate influence, 3 signifies strong influence, and 4 indicates extremely strong influence. The questionnaire design is clear and straightforward, ensuring participating experts can accurately comprehend and provide feedback.

In selecting experts, this study invited 10 specialists with extensive academic and practical backgrounds in emergency management and supply chain fields to participate in the scoring process (

Table 3. Expert Panel Member Information.

Expert	Occupation	Title
1	Doctoral Supervisor	Professor
2	Emergency Management Agency Personnel	Policy Advisor
3	Master’s Supervisor	Professor
4	Logistics Enterprise Practitioner	Supply Chain Product Manager
5	Corporate Operations Manager	Supply Chain Operations Consultant
6	Emergency Logistics Supply Chain Practitioner	Legal Counsel
7	Emergency Field Researcher and Educator	Associate Professor
8	Emergency Management Research Institute Member	Researcher
9	Logistics Engineering Researcher and Educator	Director of Research Institute
10	Emergency Management Practitioner	Technical Engineer

). This sample size aligns with the typical panel size (usually 8–15 members) in similar research methodologies, ensuring information saturation while balancing data collection feasibility and expert opinion representativeness. All experts possessed an average professional experience exceeding 15 years, with research domains highly relevant to this topic. The panel comprised three scholars specializing in emergency logistics policy research, four researchers focused on supply chain resilience modeling, and three experts with extensive practical experience in intelligent logistics technology applications or supply chain risk management. The expert composition encompassed diverse perspectives, including university professors, senior corporate executives, policymakers, and senior technical engineers. This ensured the study comprehensively reflected the practical realities of emergency logistics supply chains across theoretical, policy, operational, and technological dimensions, enhancing the universality and practical reference value of the research conclusions.

To enhance the scientific rigor and objectivity of the scoring process while minimizing subjective bias, this study implemented the following steps: First, prior to formal scoring, all experts received a standardized research background introduction, a four-dimensional framework explanation, and detailed scoring criteria to ensure consistent understanding of the research objectives and measurement scale. Second, Kendall’s coefficient of concordance and Cronbach’s alpha were used to assess the consistency and reliability of expert scoring, respectively. Results indicated an α coefficient of 0.85 (>0.8), demonstrating strong internal consistency among expert ratings. During data processing, to mitigate potential subjective bias from individual extreme scores, the highest and lowest ratings within each group were excluded. The arithmetic mean of the remaining valid scores served as the final quantitative value, thereby further enhancing data robustness and objectivity.

It should be noted that while expert-based scoring evaluation methods are widely adopted in complex systems and emerging field research, they still exhibit certain limitations. These include susceptibility to individual experts’ experience, cognitive biases, and disciplinary constraints. To mitigate such biases, this study emphasizes background diversity in expert selection and implements quality control through standardized instructions and statistical verification before and after scoring.

4.2. Analysis of Fuzzy DEMATEL Results

4.2.1. Centrality and Causality Analysis

By applying deblurring through Equations (1)–(6), the direct influence matrix is obtained as shown in

Table 4. Direct Impact Matrix.

Factor	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10	S11	S12	S13	S14	S15	S16	S17
S1	0	0.2667	0.5	0.5	0.5	0.7333	0.5	0.5	0.7125	0.5	0.7333	0.5	0.2667	0.2667	0.0333	0.2667	0.2667
S2	0.7333	0	0.5	0.7333	0.5	0.5	0.5	0.7333	0.4875	0.5	0.7333	0.5	0.5	0.2667	0.5	0.2667	0.5
S3	0.5	0.5	0	0.2667	0.5	0.7333	0.7333	0.5	0.7125	0.5	0.5	0.2667	0.2667	0.0333	0.2667	0.2667	0.2667
S4	0.5	0.5	0.2667	0	0.5	0.7333	0.7333	0.7333	0.4875	0.2667	0.7333	0.2667	0.2667	0.0333	0.0333	0.2667	0.2667
S5	0.7333	0.5	0.5	0.5	0	0.7333	0.7333	0.7333	0.7125	0.5	0.7333	0.7333	0.5	0.5	0.2667	0.2667	0.2667
S6	0.2667	0.0333	0.2667	0.2667	0.0333	0	0.7333	0.2667	0.7125	0.2667	0.7333	0.0333	0.0333	0.0333	0.0333	0.0333	0.0333
S7	0.2667	0.0333	0.2667	0.2667	0.0333	0.7333	0	0.2667	0.4875	0.2667	0.5	0.0333	0.0333	0.0333	0.0333	0.0333	0.0333
S8	0.5	0.5	0.5	0.5	0.2667	0.7333	0.7333	0	0.7125	0.5	0.7333	0.2667	0.2667	0.0333	0.2667	0.2667	0.2667
S9	0.2667	0.0333	0.2667	0.2667	0.0333	0.5	0.5	0.2667	0	0.2667	0.5	0.0333	0.0333	0.0333	0.0333	0.0333	0.0333
S10	0.7333	0.2667	0.5	0.2667	0.5	0.7333	0.5	0.7333	0.7125	0	0.7333	0.5	0.2667	0.2667	0.2667	0.2667	0.2667
S11	0.2667	0.0333	0.2667	0.2667	0.0333	0.7333	0.7333	0.2667	0.4875	0.2667	0	0.0333	0.0333	0.0333	0.0333	0.0333	0.0333
S12	0.7333	0.5	0.5	0.5	0.7333	0.7333	0.5	0.5	0.4875	0.7333	0.7333	0	0.2667	0.5	0.5	0.2667	0.2667
S13	0.7333	0.7333	0.5	0.5	0.5	0.7333	0.5	0.5	0.2625	0.7333	0.7333	0.5	0	0.5	0.5	0.5	0.5
S14	0.7333	0.5	0.5	0.5	0.7333	0.7333	0.5	0.5	0.4875	0.5	0.7333	0.7333	0.5	0	0.7333	0.5	0.5
S15	0.7333	0.5	0.7333	0.5	0.5	0.7333	0.7333	0.7333	0.4875	0.5	0.7333	0.7333	0.5	0.7333	0	0.5	0.5
S16	0.5	0.5	0.5	0.5	0.9667	0.7333	0.5	0.5	1	0.7333	0.7333	0.5	0.7333	0.7333	0.7333	0	0.9667
S17	0.7333	0.7333	0.7333	0.5	0.9667	0.5	0.7333	0.7333	0.4875	0.7333	0.7333	0.7333	0.7333	0.7333	0.7333	0.9667	0

.

Based on Equations (7) and (8), the comprehensive influence matrix is calculated. Using Equation (9), the influence degree $D_i$, affected degree $E_i$, centrality $M_i$, and causality degree $R_i$ for each influencing factor are computed. The attribute information for these four key indicators is presented in

Table 5. Influence Degree, Affected Degree, Centrality, and Causality Degree for Each Factor.

Factor	Influence Degree	Affected Degree	Centrality	Causality	Ranking	Attributes
S1	1.4018	1.8183	3.2201	−0.4165	3	Resulting Factors
S2	1.777	1.204	2.9809	0.573	10	Causal Factors
S3	1.3582	1.5111	2.8693	−0.1529	14	Resulting Factors
S4	1.2969	1.4259	2.7228	−0.129	16	Resulting Factors
S5	1.8087	1.4103	3.2189	0.3984	4	Causal Factors
S6	0.6491	2.3551	3.0042	−1.706	8	Resulting Factors
S7	0.5894	2.1352	2.7246	−1.5457	15	Resulting Factors
S8	1.3798	1.7458	3.1256	−0.366	5	Resulting Factors
S9	0.559	2.0444	2.6033	−1.4854	17	Resulting Factors
S10	1.51	1.5934	3.1035	−0.0834	6	Resulting Factors
S11	0.6204	2.3268	2.9472	−1.7065	11	Resulting Factors
S12	1.7552	1.2423	2.9975	0.5129	9	Causal Factors
S13	1.9151	1.009	2.9241	0.9061	12	Causal Factors
S14	2.0097	0.898	2.9077	1.1116	13	Causal Factors
S15	2.0768	0.9406	3.0175	1.1362	7	Causal Factors
S16	2.3312	0.92	3.2512	1.4111	2	Causal Factors
S17	2.5033	0.9613	3.4646	1.542	1	Causal Factors

.

Based on Table 5, a causal diagram (Figure 4) was constructed to illustrate the factors influencing the resilience of emergency logistics supply chains. This diagram clearly depicts the distribution of each influencing factor and quantifies their respective impact levels on the resilience of emergency logistics supply chains.

Higher centrality among factors influencing the resilience of emergency logistics supply chains indicates greater importance. As demonstrated in Table 5 and Figure 4, S1 risk forecasting and prediction, S5 completeness of the emergency plan, S16 legal and regulatory support, and S17 funding guarantee are identified as the top four factors affecting centrality. These variables are critical to the resilience of emergency logistics supply chains and should receive prioritized attention during supply chain operations. At the Development Capacity level, S16 and S17 exhibit higher centrality, indicating that a highly resilient emergency logistics supply chain system must be built upon a foundation of adequate, diversified funding arrangements combined with a robust and clear legal and regulatory framework. Additionally, cultivating outstanding talent and actively pursuing learning and innovation also positively impact the resilience of emergency logistics supply chains. At the Resistance level, S1 and S2 exert a strong influence on emergency logistics supply chain resilience, underscoring the critical importance of strengthening infrastructure development and risk warning capabilities to prepare for sudden crises and incidents. At the adaptability level, S12 and S13 are the most crucial factors. Conducting simulation exercises and integrating new technologies helps accumulate experience and wisdom, driving the enhancement of emergency logistics supply chain resilience.

Causality serves as an indicator in which a value greater than zero denotes a causal factor, whereas a value less than zero indicates a resultant factor. Table 5 and Figure 4 demonstrate that, among the factors influencing emergency logistics supply chain resilience, eight are identified as causal factors and nine as resultant factors. Of the causal factors, S14 learning and innovating, S15 talent cultivation, S16 legal and regulatory support, and S17 funding guarantee possess the highest causality values and exert the most significant influence on other factors.

S16 and S17 function at the macro level, offering strategic guidance for emergency logistics supply chain resilience. These factors exert significant influence on other variables and demonstrate relative resistance to external disruptions. Consequently, S16 and S17 are identified as fundamental determinants of emergency logistics supply chain resilience. In contrast, S6 decision making and responsiveness, S7 transportation capacity, S9 social stability, and S11 dynamic adjustment capability display lower levels of causality and are more susceptible to external influences. These factors act as motivators that directly affect emergency logistics supply chain resilience, highlighting the need for proactive management and improvement.

4.2.2. Influence and Receptivity Analysis

Influence and receptivity are metrics for measuring the interrelated structure of multiple factors within complex systems. As shown in Figure 5, the top five factors ranked by influence are S13 integration of emerging technologies, S14 learning and innovating, S15 talent cultivation, S16 legal and regulatory support, and S17 funding guarantee, indicating these five factors exert greater influence on others. Among these, S17 funding guarantee exhibits the highest centrality, highlighting that strengthening funding guarantees is crucial for enhancing the resilience of emergency logistics supply chains.

The five factors with the highest susceptibility are S1 risk forecasting and prediction, S6 decision making and responsiveness, S7 transportation capacity, S9 social stability, and S11 dynamic adjustment capability. These factors are particularly vulnerable to the influence of other variables. Notably, S1 risk forecasting and prediction demonstrates high centrality, indicating that although early risk warning and prediction are influenced by external factors, they also significantly affect the resilience of emergency logistics supply chains.

4.3. Analysis of ISM Results

The DEMATEL method yields a comprehensive influence matrix T that quantifies the continuous causal strength between factors. However, this approach results in an overly complex system structure, making it difficult to identify critical paths. The ISMaims to construct a binary hierarchical structure, necessitating a mapping mechanism from continuous numerical values to binary relationships. This study introduces a threshold λ to eliminate redundant weak associations while retaining critical strong causal relationships, thereby guiding the construction of the ISM adjacency matrix. The value of λ is typically selected based on expert experience. To reduce the subjectivity in λ selection, this study employs a statistical approach, setting the threshold as the sum of the mean and standard deviation of the comprehensive influence matrix. Calculations yielded λ = 0.137. Based on this threshold, the adjacency matrix F was constructed using Formula (11), and the reachability matrix K was ultimately obtained through Boolean operations according to Formula (12). The adjacency matrix (

Table 6. Adjacency Matrix.

	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10	S11	S12	S13	S14	S15	S16	S17
S1	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0
S2	0	0	1	1	0	0	0	1	0	0	0	0	0	0	0	0	0
S3	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0
S4	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	0	0
S5	0	0	0	0	0	0	0	1	0	1	0	0	0	0	0	0	0
S6	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
S7	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
S8	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0
S9	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
S10	0	0	0	0	0	1	0	0	1	0	1	0	0	0	0	0	0
S11	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
S12	0	0	0	0	0	0	0	1	0	1	0	0	0	0	0	0	0
S13	1	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0
S14	1	0	0	0	1	0	0	0	0	0	0	1	0	0	0	0	0
S15	0	0	0	0	1	0	0	0	0	0	0	0	0	1	0	0	0
S16	0	0	0	0	0	0	0	0	0	0	0	0	1	0	1	0	1
S17	0	1	0	0	0	0	0	0	0	0	0	0	1	0	1	0	0

) and reachability matrix (

Table 7. Reachability Matrix.

	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10	S11	S12	S13	S14	S15	S16	S17
S1	1	0	0	0	0	0	0	1	0	0	1	0	0	0	0	0	0
S2	0	1	1	1	0	0	1	1	0	0	1	0	0	0	0	0	0
S3	0	0	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0
S4	0	0	0	1	0	0	1	0	0	0	1	0	0	0	0	0	0
S5	0	0	0	0	1	1	0	1	1	1	1	0	0	0	0	0	0
S6	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0
S7	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0
S8	0	0	0	0	0	0	0	1	0	0	1	0	0	0	0	0	0
S9	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0
S10	0	0	0	0	0	1	0	0	1	1	1	0	0	0	0	0	0
S11	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0
S12	0	0	0	0	0	1	0	1	1	1	1	1	0	0	0	0	0
S13	1	0	0	0	0	1	0	1	1	1	1	1	1	0	0	0	0
S14	1	0	1	0	1	1	1	1	1	1	1	1	0	1	0	0	0
S15	1	0	1	0	1	1	1	1	1	1	1	1	0	1	1	0	0
S16	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
S17	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	0	1

) obtained in this study are shown below.

According to the hierarchical division principle, the reachability matrix K can be hierarchically decomposed into reachable sets, prior sets, and their intersections. The set of influencing factors is shown in

Table 8. Set of Influencing Factors.

Factors	Reachable Set	Preceding Causes Collection	Intersection
S1	1, 8, 11	1, 13, 14, 15, 16, 17	1
S2	2, 3, 4, 7, 8, 11, 12	2, 12, 16, 17	2, 12
S3	3, 7	2, 3, 16, 17	3
S4	4, 7, 10, 11	2, 4, 10, 16, 17	4, 10
S5	5, 6, 8, 9, 10, 11	5, 14, 15, 16, 17	5
S6	6	5, 6, 10, 12, 13, 14, 15, 16, 17	6
S7	7	2, 3, 4, 7, 16, 17	7
S8	8, 11	1, 2, 5, 8, 12, 13, 14, 15, 16, 17	8
S9	9	2, 9, 10, 12, 13, 14, 15, 16, 17	9
S10	4, 6, 9, 10, 11	4, 5, 10, 12, 13, 14, 15, 16, 17	4, 10
S11	11	1, 2, 4, 5, 8, 10, 12, 13, 14, 15, 16, 17	11
S12	2, 6, 8, 9, 10, 11, 12	2, 12, 13, 14, 15, 16, 17	2, 12
S13	1, 6, 7, 8, 9, 10, 11, 12, 13	13, 16, 17	13
S14	1, 3, 5, 6, 7, 8, 9, 10, 11, 12, 14	14, 15, 16, 17	14
S15	1, 3, 5, 6, 7, 8, 9, 10, 11, 12, 14, 15	15, 16, 17	15
S16	1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17	16, 17	16, 17
S17	1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17	16, 17	16, 17

.

Based on the results of the element hierarchical classification in Table 8, the following final hierarchical results were obtained: L1 = {S6, S7, S9, S11}; L2 = {S3, S4, S8, S10}; L3 = {S1, S2, S5, S12}; L4 = {S13, S14, S15}; L5 = {S16, S17}. The figure illustrates the ISM hierarchical structure of emergency logistics supply chain resilience. Based on the hierarchical analysis results, a multi-level recursive structural model was constructed to determine the recursive structural relationships among system elements. As shown in Figure 6, the emergency logistics supply chain resilience system is a five-level multi-level recursive system.

The ISM facilitates the visualization of hierarchical structures and influence pathways among factors. Figure 6 illustrates the connections between factors influencing emergency logistics supply chain resilience, encompassing relationships at the same level, between adjacent levels, and across levels. In the ISM, all factors are organized into three tiers and five levels. Each factor demonstrates distinct hierarchical changes, resulting in an inverted five-tier pyramid distribution. Based on this structure, further analysis of the interactions and influence mechanisms among these factors reveals the following:

First, level 5 includes S16 legal and regulatory support and S17 funding guarantee. These factors represent the most fundamental and critical elements within the entire system, constituting the essential drivers of emergency logistics supply chain resilience. However, legal and regulatory support and funding guarantee do not directly impact emergency logistics supply chain resilience; instead, they influence resilience by affecting other factors.

Second, the seven intermediate factors—S1 Risk Forecasting and Prediction, S2 Infrastructure Construction, S5 Completeness of the Emergency Plan, S12 Simulation and Drill, S13 Integration of Emerging Technologies, S14 Learning and Innovation, and S15 Talent Cultivation—are situated in the third and fourth layers of the model. These factors are shaped by essential factors and, in turn, influence surface-level factors. Specifically, risk forecasting and prediction, infrastructure construction, and completeness of the emergency plan are components of the emergency logistics supply chain system. Simulation and drill and integration of emerging technologies are part of the technical system, while learning and innovation and talent cultivation are associated with the management system. Although these intermediate factors do not directly affect specific elements of the emergency logistics supply chain, they indirectly enhance its resilience by influencing surface-level factors. Consequently, it is necessary to fully integrate these factors into the emergency logistics supply chain system and implement targeted measures to strengthen resilience.

Third, S3 equipment maintenance, S4 natural environment compatibility, S6 decision making and responsiveness, S7 transportation capacity, S8 logistical maintenance and storage management, S9 social stability, S10 data sharing and coordination, and S11 dynamic adjustment capability are situated within the first and second layers of the model. These factors serve as direct drivers of emergency logistics supply chain resilience, although they are also shaped by intermediate and root factors. Equipment maintenance and natural environment compatibility are classified under the Resistance dimension. Decision making and responsiveness, transportation capacity, logistical maintenance and storage management, and social stability are associated with the responsiveness dimension. Data sharing and coordination, as well as dynamic adjustment capability, are components of the adaptability dimension. Collectively, these factors exert a direct influence on the resilience of emergency logistics supply chains by affecting their resistance, recovery, and adaptability. Consequently, it is essential to prioritize and optimize these factors to strengthen the resilience of emergency logistics supply chains.

4.4. MICMAC Results Analysis

Based on the reachability matrix K, the driving force and dependency degree of each influencing factor were calculated according to Formula (14), with the results shown in

Table 9. Driving Forces and Dependencies of Factors.

Factors	Driving Force	Dependency
S1	3	6
S2	6	3
S3	2	6
S4	3	4
S5	6	5
S6	1	9
S7	1	8
S8	2	10
S9	1	9
S10	4	8
S11	1	13
S12	6	6
S13	8	3
S14	11	4
S15	12	3
S16	17	1
S17	16	2

.

Table 9 categorizes the factors into four distinct quadrants: autonomous factors, dependent factors, associated factors, and independent factors, as shown in Figure 7. Autonomous factors are characterized by both weak driving force and weak dependency. Dependent factors exhibit strong dependency but weak driving force. Associated factors are defined by both strong driving force and strong dependency. Independent factors possess strong driving force but weak dependency.

Dependency factors S6 decision making and responsiveness, S7 transportation capacity, S8 logistical maintenance and storage management, S9 social stability, S10 data sharing and coordination, and S11 dynamic adjustment capability are situated in the second quadrant. These factors demonstrate high dependency and low driving force, rendering them more susceptible to influence from other factors. Positioned at the direct level of the ISM, their effectiveness can be improved by strengthening the indicators of other influencing factors.

Autonomous factors S1 risk forecasting and prediction, S2 infrastructure construction, S3 equipment maintenance, S4 natural environment compatibility, S5 completeness of the emergency plan, S12 simulation and drill, and S13 integration of emerging technologies are positioned in the third quadrant. These factors exhibit low dependency and low driving force, making them relatively independent and easier to control. They are currently the focal point and starting point for enhancing the resilience of emergency logistics supply chains. Simultaneously, these factors occupy the middle layer of the ISM, serving as a bridge connecting the top and bottom layers.

The independent factors S14 Learning and innovation, S15 Talent cultivation, S16 Legal and regulatory support, and S17 Funding guarantee are positioned in the fourth quadrant. These factors exhibit low dependency and high driving force, making them less susceptible to influence from other variables and placing them at the bottom of the model. Additionally, they demonstrate high centrality within the DEMATEL model.

In summary, a comparison of MICMAC analysis results with the DEMATEL-ISMsupports the accuracy and validity of the modeling approach employed in this study.

5. Discussion

Emergency logistics supply chains are dynamic collaborative networks rapidly established to respond to sudden incidents. Their core objective is to ensure the precise, efficient, and continuous supply of critical materials and services, thereby maximizing mitigation of crisis impacts and safeguarding social order and life safety. Supply chain resilience is widely recognized as the capability of a supply chain system to maintain core functions and achieve sustainable development through proactive prevention, rapid adaptation, and recovery when facing sudden shocks.

This study achieves two significant innovative breakthroughs in emergency logistics supply chain resilience. First, it validates the pivotal role of the “Development Capacity” dimension within emergency logistics supply chain resilience, theoretically elucidating how emergency logistics systems achieve a paradigm shift from passive defense to proactive evolution through “double-loop learning” amid extreme uncertainty. This finding fills a gap in existing theories regarding how systems can benefit from disasters, constituting a significant theoretical contribution distinguishing this work from prior research. Second, unlike previous studies emphasizing decision-making and response [64,71] or data sharing [65] as primary factors influencing supply chain resilience, emergency logistics supply chains, centered on achieving humanitarian relief, operate in environments characterized by sudden surges in demand and high uncertainty. Therefore, by comparing and learning from resilience factors in other supply chains—such as cross-border e-commerce [42], prefabricated construction [58], and agriculture [61]—it identifies more targeted influencing factors, including funding guarantee, simulation and drill, natural environment compatibility, and social stability. The indicator system for influencing factors constructed in this paper not only validates the universality of previous research but also provides a specialized analytical framework for studying the resilience of emergency logistics supply chains.

The DEMATEL-ISM modeling approach is utilized to analyze factors influencing the resilience of emergency logistics supply chains, resulting in a three-tier, five-level structural diagram. Influencing factors are classified as direct, transitional, or fundamental. The first-level factors include decision making and responsiveness, transportation capacity, social stability, and dynamic adjustment capability, which constitute the most direct causes of emergency logistics supply chain resilience. These findings suggest that emergency management should involve advanced preparation of multiple transportation plans and supplier options, as well as the maintenance of basic social order to ensure stable supply chain operations. This conclusion aligns with previous research emphasizing the significance of multi-supplier strategies [72].

The second layer comprises secondary direct factors: equipment maintenance, natural environment compatibility, logistical maintenance and storage management, and data sharing and coordination. This tier emphasizes strengthening routine maintenance of equipment and facilities, optimizing warehouse management, and ensuring real-time data sharing post-incident to enhance coordination efficiency and prevent information silos. Additionally, supply chain node layouts must align with local natural environments, a conclusion supported by research from Turken et al. [73].

The third and fourth layers represent transitional factors, with some focusing on preemptive prevention—such as risk forecasting and prediction, infrastructure construction, and completeness of emergency plans—enhancing overall supply chain resilience through improved capabilities in these areas. Others concentrate on strengthening and upgrading emergency logistics supply chains, including the integration of emerging technologies, simulation and drills, learning and innovation, and talent cultivation. This necessitates that governments and relevant organizations establish post-disaster review and knowledge management systems to transform case experiences into standardized procedures. Concurrently, they should strengthen emergency logistics supply chain-related curricula and interdisciplinary talent development in universities and enterprises while actively integrating new technologies like AI, IoT, and blockchain to enhance the system’s adaptability and continuous learning capabilities. It is noteworthy that despite the ISMdemonstrating a clear bottom-up driving pathway, the actual operation of emergency logistics systems exhibits significant cross-level feedback mechanisms. Although positioned at the middle layer of the model, “Learning and innovating” and “Talent cultivation” serve as pivotal connectors between underlying resources and surface-level capabilities. Meanwhile, many existing studies regard infrastructure and coordination capacity as the primary drivers of resilience enhancement. While acknowledging their importance, this study’s findings offer a different perspective through MICMAC analysis: within emergency logistics supply chain systems, these factors manifest as “strongly dependent” intermediate variables or outcome variables rather than root drivers. This suggests that infrastructure investments alone may fail to translate into effective system resilience if detached from overarching strategic planning (e.g., S1, S5) and foundational institutional safeguards (S16, S17). This paradigm shift represents a unique insight derived from this study, grounded in the “government-led, multi-stakeholder collaborative” nature of emergency logistics. This discovery fills a gap in existing literature regarding the mechanisms through which emergency systems learn from disasters, demonstrating that true resilience extends beyond recovery to encompass systemic development and evolution.

The fifth layer encompasses legal and regulatory support and funding guarantees. These factors permeate and underpin the entire emergency logistics supply chain development process, indicating that the state should incorporate emergency logistics supply chains into long-term development plans and improve diversified funding guarantee systems to prevent disruptions in rescue resources due to policy or funding fluctuations. This outcome profoundly reflects the unique constraints imposed by the suddenness and high uncertainty of emergency logistics on resilience development. First, during the initial stages of a disaster, conventional market-based financing mechanisms and routine administrative approval processes often fail due to delayed responses. At this critical juncture, legal authorizations and reserve funds under emergency conditions ensure that supply chains retain lawful operational authority and financial capacity even when conventional order collapses. Second, to mitigate resource misallocation risks stemming from high uncertainty, relying solely on warehoused inventory proves costly in transportation and challenging to maintain. Adequate financial safeguards actually empower the system to flexibly allocate societal resources within uncertain environments. Thus, these two factors constitute institutional redundancy for coping with extreme conditions, whose operational mechanisms far exceed those of conventional resource inputs.

From a sustainable development perspective, this outcome aligns closely with institutional theory. Emergency logistics supply chains exhibit pronounced quasi-public goods characteristics, and their sustainability cannot be maintained solely through market mechanisms. S16 and S17, as the most potent root factors, effectively constitute the system’s institutional resilience. A robust legal framework ensures the legitimacy and social equity of emergency operations, while adequate financial safeguards provide the material foundation for achieving the Sustainable Development Goal of “building resilient infrastructure.” Only by strengthening these two foundational elements can emergency logistics systems break free from the vicious cycle of “disaster-reconstruction-recurring disaster” and transition toward a secure, sustainable development model.

Based on the aforementioned findings and analysis, this study identifies discrepancies with existing research conclusions:

First, unlike studies examining commercial supply chains, this research identifies unique key factors through comparative analysis. For instance, Sathyan et al. [64] identified supplier relationship management and process flexibility as core drivers in their automotive supply chain resilience study; similarly, Liu et al. [42] emphasized logistics infrastructure as a critical independent driver in their cross-border e-commerce research. In stark contrast, our findings reveal a fundamental divergence: within emergency logistics contexts, regulatory support (S16) and financial safeguards (S17) emerge as absolute root drivers (independent factors), while operational elements like infrastructure and flexibility are categorized as dependent or intermediate variables. The theoretical rationale behind this divergence lies in the failure of market mechanisms during disasters. Commercial supply chains rely on profit motives to spontaneously drive resource allocation, whereas emergency logistics constitutes a quasi-public good. In extreme disaster scenarios, physical capabilities (infrastructure and transportation) cannot be effectively activated without top-down authorization provided by legal frameworks and non-market liquidity supplied by dedicated funds. Thus, the dominance of institutional factors over operational factors constitutes a unique insight of this study, offering a distinct perspective to the efficiency-first logic prevalent in commercial supply chain literature.

Second, this study offers fresh perspectives on the attributes of intermediate factors. Previous research, such as Brandon-Jones et al. [20], typically treats information sharing and visibility as primary precursors to resilience. However, our MICMAC analysis classifies data sharing and coordination (S10) as dependent variables (second quadrant), highly reliant on other factors. This finding diverges from the technological determinism prevalent in existing literature. It indicates that in chaotic emergency response settings, data silos represent not merely technical failures but institutional failures. Without Learning and Innovation (S14) and Talent Development (S15)—identified in our model as strong drivers—even advanced information platforms cannot translate data into effective decisions. This highlights that for emergency logistics, learning and talent serve as critical bridges to activate technological resources, a mechanism more pronounced than in stable commercial environments.

Finally, the findings of this study were validated through mixed-methods triangulation. The factor attributes identified in the MICMAC analysis were fully consistent with the DEMATEL-ISM analysis, and this methodological cross-validation enhances the reliability of the conclusions. More importantly, the four-dimensional resilience framework constructed in this study establishes a bridge from resilience to sustainability. Traditional perspectives often view resilience as short-term stress resistance. This study demonstrates that by enhancing Development Capacity, emergency logistics supply chains can transform one-time disaster responses into long-term organizational assets. This mechanism not only improves immediate disaster relief efficiency but also directly contributes to the long-term sustainable development of society and the environment by reducing vulnerability to future disasters.

6. Conclusions and Implication

6.1. Conclusions

To further deepen the theoretical framework of emergency logistics supply chain resilience and enhance its risk resistance and crisis prevention capabilities, this paper constructs a hybrid fuzzy DEMATEL-ISM-MICMAC analysis method. This method provides a comprehensive analytical framework for identifying key factors influencing emergency logistics supply chain resilience and visually presents the complex systemic relationships among these factors. The main research conclusions are as follows:

(1) Literature analysis and case study methods, in conjunction with the four resilience indicators of resistance, responsiveness, adaptability, and development capacity, were used to establish an indicator system suitable for assessing factors influencing emergency logistics supply chain resilience.

(2) Analysis of organizational resilience factors using the fuzzy DEMATEL method indicates that infrastructure construction, completeness of the emergency plan, integration of emerging technologies, learning and innovation, talent cultivation, legal and regulatory support, and funding guarantee are central to the emergency logistics supply chain resilience system and exert the most significant influence. Consequently, these critical factors should be prioritized when enhancing emergency logistics supply chain resilience, particularly during its dynamic evolution.

(3) ISM analysis demonstrates that legal and regulatory support, as well as funding guarantees, are fundamental factors influencing the resilience of emergency logistics supply chains. Adequate and diversified funding arrangements, combined with robust and clear legal and regulatory frameworks, systematically enhance supply chain resilience at multiple levels and dimensions, thereby ensuring efficient emergency logistics operations.

(4) MICMAC analysis further identifies learning and innovation, talent cultivation, legal and regulatory support, and funding guarantee as foundational factors with strong driving power and weak dependency. These factors propel the development of other elements within the system. The study’s conclusions offer theoretical foundations and practical references for optimizing the resilience of emergency logistics supply chains.

6.2. Management Insights and Implementation Pathways

Based on the analysis results of the hybrid fuzzy DEMATEL-ISM-MICMAC model, this study reveals that the resilience of emergency logistics supply chains constitutes a multi-level hierarchical system where deep-level institutional safeguards drive mid-level capacity building, which in turn determines surface-level responsiveness. To effectively enhance supply chain resilience, this paper proposes a multi-agent collaborative optimization strategy with priority ranking, integrated with the ISM hierarchical structure.

The ISM indicates that legal support and financial safeguards occupy the foundational layer as the system’s root drivers, commanding the highest priority. Without these fundamentals, other technical or managerial measures lack essential underpinning.

Governments should elevate emergency logistics supply chain resilience to a national security strategy. The primary task is to refine the legal framework, clarifying compensation standards and liability exemptions for government–enterprise coordination to eliminate enterprises’ legal concerns about emergency participation. Second, diversified funding pools must be established, increasing dedicated emergency allocations and encouraging multi-stakeholder co-investment. Policy incentives should motivate stakeholders to enhance emergency supply chain capabilities.

Factors at the model’s intermediate layer—learning and innovation, talent development, and emerging technology integration—serve as pivotal connectors. This phase aims to translate institutional dividends into tangible core competencies, emphasizing the shift from reactive defense to proactive evolution.

Universities and research institutions should establish industry-academia-research bases for emergency logistics targeting S15 and S14, prioritizing the cultivation of multidisciplinary talent with big data analytics and crisis management capabilities. They should also assist in building post-disaster debriefing and knowledge management repositories to ensure every disaster response experience is translated into standardized operational procedures.

Technology providers and platform enterprises should focus on S13 and S12, leveraging digital twin technology to construct emergency simulation systems. Through routine virtual exercises, they can not only cost-effectively test contingency plan completeness but also establish cross-departmental collaboration mechanisms before actual operations, enhancing the intelligence of infrastructure and early risk warning systems.

Factors at the model’s surface layer directly embody resilience and represent the ultimate outcome of the first two stages’ development. This phase emphasizes rapid response and precise execution during actual operations.

Emergency command centers must leverage mid-level data sharing capabilities to establish streamlined decision-making mechanisms. During crises, they should fully utilize the financial and legal authority granted at the grassroots level to rapidly mobilize social transport capacity and achieve dynamic cross-regional resource allocation.

At the governmental level, investment in public infrastructure such as emergency logistics hubs and communication networks must be strengthened. Concurrently, public safety education campaigns should be intensified, with timely dissemination of accurate and transparent information to maintain social order and stability.

Logistics enterprises and social organizations, as executors of transportation, warehousing management, and logistical support, should prioritize routine equipment maintenance and environmental adaptation upgrades. During disasters, they should leverage pre-established collaborative networks to dynamically adjust delivery routes based on real-time disaster data, ensuring “last-mile” material accessibility to uphold social stability.

In summary, policymakers and administrators should follow a path grounded in institutional frameworks and funding, supported by technological advancements and talent development, and implemented through operational coordination. This causally layered governance logic will propel emergency logistics supply chains from relying on ad hoc responses to achieving a resilient leap with evolutionary capabilities.

6.3. Limitations and Future Research

This research makes valuable contributions to advancing the strengthening and sustainable development of emergency logistics supply chain resilience, though room for improvement remains.

First, the accuracy of indicator weighting directly impacts the quality of the entire resilience assessment process. Although this study implemented multiple measures to enhance result objectivity, it must be acknowledged that expert-based evaluation methods retain inherent limitations. Even experienced experts may be constrained by their specific professional backgrounds or recent experiences, potentially introducing subjective elements into the outcomes. Therefore, subsequent work should explore more objective and precise data processing methods to enhance the reliability of the data, thereby laying a solid foundation for in-depth analysis of such issues.

Second, since extreme disasters are low-probability, high-impact events lacking sufficient big data to support empirical regression analysis, relying on expert knowledge remains the most effective approach for deconstructing such complex systems. Future research could further validate and refine the conclusions of this study by expanding sample sizes or integrating quantitative methods such as system dynamics simulations, thereby better addressing the challenges posed by increasingly complex emergency situations.

Finally, this study did not fully account for the moderating effects arising from regional variations in disaster types. The expert panel primarily drew from a Chinese context and focused on generic emergency logistics scenarios. Given significant differences in geopolitical environments, infrastructure levels, and disaster types across nations, the universality of the model constructed in this study may be limited in cross-cultural or specific extreme scenarios. Nevertheless, the proposed four-dimensional framework itself possesses inherent universality and can serve as a common yardstick for assessing supply chain resilience across diverse institutional systems. Future research could further refine the model to enhance its applicability in specific extreme scenarios.