搜索结果：Risk analysis : an official publication of the Society for Risk Analysis

Unmanned aerial vehicles (UAVs) are increasingly considered for urban logistics, including the delivery of critical medical supplies. However, their safe operation in dense urban environments requires systematic risk analysis that can account for both technological uncertainties and human factors. Conventional failure mode and effects analysis (FMEA) offers a structured way to identify and prioritize risks but is limited by its use of deterministic evaluations and the absence of mechanisms to handle uncertainty or disagreement among experts. To address these limitations, this study develops an integrated cloud-based FMEA framework to UAV logistics operations. The method begins with expert assessments expressed through linguistic terms, which are transformed into cloud models to represent uncertainty more faithfully. A consensus-reaching process is then applied to align divergent expert opinions and establish an aggregated evaluation matrix. To determine the relative importance of risk factors, a hybrid weighting scheme is employed that combines subjective judgments through cloud step-wise weight assessment ratio analysis (SWARA) and objective information through cloud criteria importance through intercriteria correlation (CRITIC). These weights are subsequently integrated into a cloud technique for order similarity to ideal solution (TOPSIS) procedure that produces a transparent ranking of failure modes. The framework is applied to a case study of medical supply delivery UAVs, where it identifies the most critical risks and validates the prioritization through sensitivity and comparative analyses. The results demonstrate that the proposed approach provides a robust and practical decision-support tool for enhancing safety in urban UAV logistics.

The lack of a dose-response model for Clostridioides difficile, attributed to insufficient human data, challenges effective C. difficile infection (CDI) risk management. This study presents a novel murine-derived dose-response model for C. difficile, developed from experimental data that closely mirrors human CDI, advancing the field of quantitative microbial risk assessment (QMRA) and the strategic management of this pathogen. Animal dose-response datasets for C. difficile were evaluated against established criteria, and a mouse model of C. difficile-associated colitis that mirrors critical aspects of CDI in human beings was selected as the most appropriate dataset for analysis. Using maximum likelihood estimation in R, these data were fitted to the beta-Poisson and exponential models and assessed through Akaike information criterion, Bayesian information criterion, likelihood tests, and a sensitivity analysis. The beta-Poisson model was identified as the best fit (α = 0.56 and N50 = 2871.56) with estimated mean doses for 10% and 50% infection rates of 505 and 3994 CFU, respectively. This model was then applied in a QMRA framework to assess CDI risk associated with commonly encountered clinical surfaces based on published surface contamination data and healthcare worker contact rates. Annual risk estimates from the model's application suggest that, out of every 100,000 healthcare workers exposed in clinical settings, approximately 33 CDI cases could result from contact with bed rails, 12 from computer keyboards, and fewer than 1 from door handles. By providing a formally derived dose-response relationship for C. difficile, this study offers a starting point for future QMRA research on this pathogen and supports the development of more detailed, context-specific risk assessments for C. difficile.

The concept of risk is used across disciplinary boundaries by researchers as well as practitioners. In this paper, we aim to explore the implications of such a use for the development of the theory and practice of risk analysis, understood broadly as covering risk assessment, risk perception, risk communication, risk management, and policy on risk. As an analytical frame, we consider risk as a "boundary object" around which different communities of practice organize their interconnections. One key implication for theory is that the way risk is conceptualized needs to balance being adaptable to disciplinary needs, yet retaining a common identity across disciplines, and we find that there are perspectives that, at least conceptually, accommodate risk functioning as a boundary object. Practitioners should expect to find that different communities of practice assign somewhat different meanings to the concept of "risk." Such inconsistencies should be taken as an opportunity to explore and learn rather than an occasion for policing and eradicating deviant discourses. Based on our findings, we outline a research agenda for advancing the understanding of risk as a boundary object, including empirical studies of whether this is actually taking place.

The world is constantly changing, yet a risk assessment is based on the knowledge available at one point in time. There will therefore be a gap between the range of possibilities known or conceivable to the assessor at that time and all the possibilities that could occur over infinite time. This will leave the door open to surprises. Anticipating surprises, therefore, requires the narrowing of this knowledge gap. We outline an approach to narrowing it that transforms risk assessors' small-world representations into a large-world representation by configuring them into a small-world network. Small-world representations are individuals' partial and subjective perspectives on an aspect of reality. What we call a large-world representation integrates these small-world representations. This transformation narrows the knowledge gap by integrating dispersed knowledge about, and intersecting alternative framings of, a focal risk, thereby dynamically updating the knowledge landscape underpinning its assessment. We use the 9/11 attack as an example. That surprise resulted from a failure to intersect the frames "suicide attack" and "hijacking," meaning that the possibility of a "suicide hijacking" went unconsidered. Configuring risk assessors into a small-world network would have increased the chance that these two frames would intersect in a risk assessment, thereby anticipating this surprise outcome. In sum, the approach we outline operationalizes the recently extended (C, U) risk-assessment framework. It increases the chance that surprises are anticipated by enabling risk assessors to see as an integral whole what they could otherwise see only fragmentarily.

Emergency response performance assessment (ERPA) is critical for improving emergency management by supporting proactive risk identification and post event improvement. Existing ERPA approaches based on conceptual frameworks or system effectiveness seldom capture operational mechanisms, the decision-makers' bounded rationality, and the joint effects of procedure dependency and organizational collaboration. To address these limitations, this study develops a dynamic metanetwork-based risk analysis framework grounded in prospect theory that emphasizes multisubjective interactions within emergency response systems and distorted performance perceptions of decision-makers. The emergency response system is represented as a dynamic metanetwork with four node types, including procedure, organization, resource, and information, connected through seven subnetworks that describe system operations. Satisfaction indicators are defined at the single-procedure level and transformed using S-shaped timeliness functions to reflect bounded rationality under prospect theory, then extended to multi-procedure settings to capture procedural dependency. Key forms of organizational collaboration are identified, including procedure hand-offs, co-assignments, and resource negotiations, and collaboration indicators are constructed to measure alignment between organizational networks and these collaboration forms. Procedure timeliness and organizational collaboration are finally integrated through a fuzzy affiliation function to account for external uncertainty. The proposed framework is illustrated through a risk ERPA of the Manchester Arena attack, with sensitivity and comparative analysis demonstrating its effectiveness in complex settings.

Bioaerosol emission characteristics from municipal solid waste (MSW) transfer points pose significant health risks to workers and residents. However, studies on its computable evaluation are rare. This study examines the emission characteristics of Staphylococcus aureus and Escherichia coli bioaerosols at various sampling points in and around the MSW transfer point in summer and autumn. Monte Carlo simulation-based quantitative microbial risk assessment was used to estimate the annual probability of infection (P(a)inf) and disease burden (DB) for worker groups (waste collector and operator) and pedestrians with or without personal protection equipment (PPE). Furthermore, sensitivity analysis identified the contribution of input variables to DB variance and rank correlation coefficients. The results show that bioaerosol concentrations at the control room were 1.19-1.34 times lower than the center point but 1.18-1.56 times higher than those at the footpath. The P(a)inf of E. coli and S. aureus bioaerosol for all exposure scenarios in the summer was 1.66-2.18 and 1.45-1.63 times, respectively, higher than that in the autumn. For all exposure populations, the DB with PPE was approximately one order of magnitude lower than that without PPE. Exposure concentration to the DB without PPE was the first predominant input parameter for all exposure scenarios. Sensitivity analysis recommends using PPE as a key mitigation strategy to manage bioaerosol health risks during different seasons and worker roles. This research delivered novel data and provided guidelines for controlling the bioaerosol emission health risks from MSW transfer points.

The proliferation of online misinformation poses severe societal risks, including public health crises and political polarization, with emotional manipulation serving as a key driver of its spread. However, traditional metrics fail to capture the nuanced role of emotions and coordinated amplification in misinformation networks. To address this gap, this study examined emotional contagion dynamics in misinformation diffusion across heterogeneous actor networks on social media. It employed machine learning-based actor classification, Exponential Random Graph Models (ERGM), and time-series analysis on retweet, reply, and quote networks. ERGM disentangled emotion-specific sender and receiver effects and temporal lagged analyses traced dynamic emotional propagation across actor categories. Key findings revealed that anger and sadness exhibited sender-dominated spread in retweet networks but receiver-driven dynamics in replies. Moreover, astroturfing actors strategically gatekept emotional flows by occupying high-betweenness positions to amplify polarization. Triadic closure effects further strengthened in-group emotional reinforcement. Taken together, these results demonstrate that emotional manipulation in misinformation ecosystems amplifies digital information risks, with astroturfing actors boosting negative emotions to trigger self-reinforcing contagion cycles. Consequently, this study underscores the need for emotion-aware risk assessment tools and context-sensitive moderation strategies. By pioneering a network-sensitive framework to quantify the risks of emotional manipulation, this work offers actionable insights for platform governance and misinformation research.

From all the knowledge that would emerge as relevant to it over infinite time, a risk analysis must be based on the cross-section available at its undertaking. This creates a knowledge gap, which can lead to surprises. To address a similar problem in economic decision-making, G. L. S. Shackle developed potential surprise theory (PST). PST's focus is on decisions that take the form of crucial, self-destructive experiments, which destroy and radically remake the possibility space. Such decisions turn the kaleidic economy, the ceaseless shape-shifting of which precludes absolute foreknowledge of all its possibilities. This dynamism means that the exhaustive listing of all possibilities required to assign probabilities necessitates the use of what Shackle called a "residual hypothesis" to represent all presently unknown scenarios. Yet, there is no way to know what probability to assign to this residual. PST overcomes this problem by employing a non-probabilistic and nonadditive measure of uncertainty. PST has much to offer the uncertainty-based perspective on risk, yet proponents of that perspective have been curiously inattentive to it. This article rectifies that by (1) showing how PST overcomes the residual-hypothesis problem that is foundational to risk analysis; (2) juxtaposing PST and expected utility theory; (3) illustrating the nature of crucial experiments in risk analysis; (4) describing PST's language of possibility and its focus on surprises and extremes; and (5) discussing PST's operationalization in a risk analysis. In summary, PST can serve as a practical and theoretical cornerstone of the uncertainty-based perspective on risk.

Emerging environmental risks are often shaped not by a lack of knowledge alone, but by fragmented information across systems, disciplines, and levels of governance. This fragmentation limits the ability of local decision-makers to identify and respond effectively to rapidly developing technologies. This paper introduces a novel bow tie risk assessment framework as a practical tool for identifying and organizing these risks. By integrating engineering, environmental, and policy perspectives, the approach captures interactions across land use, water systems, and governance structures. We apply the framework to carbon capture, utilization, and storage (CCUS) development in Colorado. Analysis shows that tracking how CCS, CCU, and CCUS are defined and applied across institutions reveals gaps in accountability, coordination, and risk identification. These inconsistencies contribute to policy drift, obscure system-level impacts, and limit stakeholder engagement. The bow tie framework makes these gaps visible, drawing attention to risks that remain hidden within fragmented knowledge and governance systems. Findings from the Colorado case study indicate that current CCUS governance lacks consistent mechanisms to define, measure, and account for water use and impacts across institutions. The analysis highlights a critical gap in how water is conceptualized and measured, particularly where it may be permanently removed or altered through subsurface injection, storage, or disposal in ways that do not align with conventional distinctions between consumptive and non-consumptive use. The method provides a practical tool for local and regional governments to identify risks, engage broader stakeholders, and support coordinated, interdisciplinary, and adaptive decision-making.

The spillover of H5N1 clade 2.3.4.4b into dairy cattle has raised concerns over the safety of fluid milk. While no foodborne infection has been reported in humans, this strain has infected at least 70 people, and milk from infected cows is known to be infected by ingestion of multiple other species. Investigation into the public health threat of this outbreak is warranted. This farm-to-table quantitative microbial risk assessment (QMRA) uses stochastic models to assess the risk of human infection from consumption of raw and pasteurized fluid cow's milk from the United States supply chains. These models were parameterized with literature emerging from this outbreak and then employed to estimate the H5N1 infection risk and evaluate multiple potential interventions aimed at reducing this risk. The median (5th and 95th percentiles) probabilities of infection per 240-mL serving of pasteurized, farmstore-purchased raw, or retail-purchased raw milk were 7.66E-19 (2.39E-20, 4.02E-17), 1.56E-7 (6.67E-10, 1.28E-5), and 1.40E-7 (6.65E-10, 1.13E-05), respectively. Our results confirm that pasteurization is highly effective at reducing H5N1 infection risk. Scenario analysis revealed quantitative real-time reverse transcriptase-polymerase chain reaction (qrRT-PCR) testing of bulk tank milk to be an effective method for numerically reducing risk from raw milk. In addition, we identify knowledge gaps related to the human H5N1 dose‒response by ingestion and raw milk consumption patterns. These findings emphasize the importance of mechanistic epidemiologic models for informing public health responses amidst outbreaks with foodborne potential and highlight the need for additional research into raw milk consumption patterns to better understand this exposure pathway.

Wildfire risk mitigation on private property is central to reducing community wildfire vulnerability. Homeowners have control over many of the key factors that contribute to wildfire risk on their parcels, yet vulnerable conditions persist. One potential explanation is a misalignment between homeowners' and trained assessors' perceptions of parcel-level wildfire risk. Prior research has documented a "risk gap" in a single community wherein owners often underestimate their parcel-level wildfire risk; however, it is unclear whether such misalignments are widespread. This study replicates and expands previous research by examining the parcel-level wildfire risk gap in 38 wildland-urban interface (WUI) communities across the Western United States using paired data from household surveys and parcel-level wildfire risk assessments by trained assessors. We find that homeowners and assessors often perceive parcel characteristics differently. Homeowners may systematically mis-weight their importance, tending to underweight key controllable factors such as building materials and defensible space, leading to meaningful divergence from trained assessor ratings. Accordingly, we also find that homeowners generally underestimate overall parcel wildfire risk compared to trained assessors. These results suggest that even when homeowners recognize attributes on their parcels that contribute to wildfire risk, they may not fully grasp how much the attributes contribute to overall parcel risk, potentially failing to recognize the need to undertake meaningful mitigation actions. The findings from this study contribute to the broader natural hazards literature on expert-layperson risk perception gaps and offer insights for improving wildfire communication, education, and mitigation strategies in WUI communities.

The largest highly pathogenic avian influenza (HPAI) H5N1 outbreak to date in the United States (US) began in early 2022 and is ongoing. The analysis presented here investigates the temporal association observed between reported confirmed detections of H5N1 (cases) in commercial and non-commercial flocks at different spatial and temporal scales. Cross-correlation analysis was performed at the national level and in five states (Minnesota, South Dakota, California, Ohio, and Pennsylvania) representing approximately 60% of commercial cases at daily, weekly, and monthly frequencies. The cross-correlation results provide no support for non-commercial cases serving as a leading indicator (i.e., early warning signal) for commercial cases. In 79 counties with both commercial and non-commercial cases, approximately 65% of the first cases were in commercial flocks. Across the counties, the median arrival time of commercial cases (41 days) preceded the median arrival time of non-commercial cases (199 days). While official reported cases do not completely describe HPAI outbreak dynamics, they represent an authoritative source of information available to firms in the poultry and egg products supply chain. Based on the analysis and results presented here, non-commercial cases did not serve as a leading indicator for commercial cases in the ongoing H5N1 outbreak in the United States.

This study aims to propose integrated process mining (IPM) approach, a novel process mining method that integrates knowledge graph, and tests its performance of model generation and utility of the generated model in identifying urban cascading disaster risks. To achieve the goal, we evaluated the effectiveness of the IPM through three sets of experiments. The first experiment tests the effectiveness of the proposed method in generating reliable disaster process models using real data of eleven Chinese cities with largely varying city features. Second, we examine whether the generated model works effectively in identifying cascading disaster risks. The third experiment analyzes performance of the IPM approach under temporal and spatial constraints to understand its applicability scope using data collected from Wuhan, Yantai, and Xiamen city. The results show that (1) the IPM method can generate process models that effectively represent past disaster patterns with even incomplete process trace samples; (2) the generated process models outperform traditional methods in identifying diverse cascading disaster risks and uncovering their complex associations; and (3) performance of the proposed method is strengthened in cities with similar city features and high temporal proximity. Finally, the article provides strategic policy recommendations: (i) Leveraging multi-source media data to decode historical process patterns; (ii) enhancing process mining through domain knowledge transfer to address data constraints; and (iii) implementing spatially and temporally precise risk identification frameworks.

The assessment of accident scenarios associated with intentional attacks to chemical and process facilities has garnered the attention of institutions and practitioners because of the exacerbation of conflicts in critical contexts. In this perspective, it is important to create a framework for the integration of conventional safety approaches, dealing with unintentional events, and security science, dealing with the analysis of intentional threats. More specifically, effective protection strategies for industrial facilities require an integrated approach that combines safety and security measures. This work proposes a methodology to evaluate the cost-effectiveness of such strategies using a probabilistic approach based on Bayesian networks (BN). The methodology incorporates attack scenarios, escalation scenarios, and the probabilistic performance of integrated safety-security barriers into a cost-benefit analysis. A dedicated cost-benefit function quantifies the economic benefits of damage reduction relative to the cost of protection strategies. A demonstrative case study evaluates different protection plans, including fireproofing and video motion detection systems. The results highlight the effectiveness of integrated safety-security measures in mitigating damage probability and controlling escalation scenarios. This methodology provides a systematic tool for plant managers and practitioners to assess and compare the economic efficiency of safety and security protection plans. It bridges key gaps in the integration of safety and security analyses, offering insights into the protection of critical infrastructures.

Risk communication can support people in making timely, well-informed behavioral decisions during extreme weather events. Today's information environment has revolutionized how individuals engage with risk communication, as they see and share more information than ever before. It is not yet known how much risk communication any specific individual encounters, where it comes from, what messages it contains, and when it is received over the course of evolving extreme weather events. In this study, we provide a detailed description of how individuals encounter risk communication during extreme weather events by analyzing comprehensive records of all the content that 11 adults viewed on their smartphones (n = 162,418 screenshots collected via mobile sensing) over 5 days of the winter disaster that struck Texas in February 2021. Participants viewed substantial amounts of risk communication, which comprised, on average, 21% of their total smartphone use during the winter disaster. Most risk communication was accessed through participants' social networks, primarily through personal messaging apps (M = 33%) and social media apps (M = 22%). Risk messages contained updates, actionable advice, social support, sense-making, and recreation and humor, the specifics of which evolved over the course of the winter disaster. Taken together, our results demonstrate how individuals engage with risk communication in dynamic, heterogeneous, and socially embedded ways during extreme weather events-opening new possibilities for risk communication theory and practice.

Household transmission risks can be especially high for respiratory viral diseases and have impactful economic and public health consequences. The study objective was to estimate influenza and rhinovirus infection risks for a child sharing a bedroom with another infected child using a computational fluid dynamics (CFD) quantitative microbial risk assessment (QMRA) approach. The scenario was two recumbent children lying on twin-sized beds in a 25.5 m3 bedroom for 6 h. The infected individual released virus-laden aerosols under scenarios: Low and high severity shedding. Rhinovirus and influenza A virus (H1N1 and H3N2) dose-response curves were used to estimate infection risks. Infection risks were highest for rhinovirus, >0.99 regardless of viral shedding intensity. High severity settings yielded risks 1.5 and 0.9 log10 higher for H3N2 and H1N1 infection risks, respectively. A plume formed from the infected individuals exhaled breath to the susceptible individual's face, which may explain, in part, the high risks estimated. While settings for tidal breathing did not disrupt the formation of this plume, increased air exchange rate did. Respiratory viral infection risks can be high in shared bedroom scenarios, especially with low fresh air ventilation and alignment of furniture such that plumes can form between the faces of individuals.

Assessing the economic losses caused by earthquakes is important for disaster relief and post-disaster loss compensation. As casualty information becomes available, it supports detailed loss estimations in accurately determining damages and facilitating economic compensation. Parametric insurance has emerged as a crucial tool for mitigating risk and compensating catastrophic losses with the benefits of simplified administration and expedited payment processes. Yet, its development is impeded by basis risk, which refers to the risk that the insurance payout does not match the actual loss. This paper introduces a hybrid Bayesian network (HBN) model to construct both a rapid loss assessment model and a post-disaster loss estimation model for earthquakes. The key difference between the two models is the integration of casualty data in the post-disaster assessment model. Furthermore, an innovative dual-parameter insurance pricing model is proposed, utilizing both magnitude and epicenter intensity-defined as the damage in the most affected area of an earthquake. This study also includes a practical application, calculating insurance premiums for Dali Prefecture in Yunnan Province. The research findings demonstrate that the HBN model outperforms traditional parametric methods, multiple linear regression (MLR), and random forest (RF) regarding predictive accuracy and interpretability. Additionally, integrating casualty data further enhances the model's predictive accuracy. The insurance premium calculations based on the dual-parameter model provide practical guidance for implementation. The accuracy of the model's predictions can significantly reduce basis risk. The paper contributes to more effective disaster response strategies and risk management practices, offering valuable insights for improving earthquake resilience and preparedness.

Societies worldwide face increasingly complex and interconnected crises that challenge their capacity for resilience. Assessing which structural indicators are most strongly associated with resilience scores requires quantitative methods capable of handling interdependencies, nonlinearities, and limited sample sizes. This study applies established global sensitivity analysis tools within an empirical resilience setting characterized by correlated structural indicators and small cross-national datasets. Using 124 indicators from the Joint Research Centre Resilience Dashboards, we analyze model-based structural correlates of societal resilience across EU Member States using the Lloyd's Register Foundation World Risk Poll Resilience Index for 2021 and 2023. Following univariate Pearson correlation screening and Random Forest stability selection, Shapley-based variance decomposition is computed on a Polynomial Chaos Expansion (PCE) surrogate, enabling equitable attribution under correlated inputs. For 2023, the results indicate a concentrated attribution structure, with a small core of indicators accounting for most explained variance (R2 = 0.85). Active citizenship has the largest model-based contribution, followed by antimicrobial resistance and years of life lost due to PM2.5 exposure, suggesting strong associations with social capital and environmental health under sustained polycrisis conditions. Environmental innovation, ecological conditions, and macroeconomic buffering form a secondary tier of contributors. By contrast, the 2021 model exhibits a more diffuse attribution profile (R2 = 0.82), with resilience scores associated more evenly with active citizenship, adult competences, digital readiness, and labor-market adjustment. Leave-one-out validation suggests that these attribution patterns are not dominated by single-country exclusions, although uncertainty remains substantial in a small cross-sectional sample. Overall, the comparison indicates a shift in the fitted attribution structure from broader recovery-oriented human-capital and digital indicators in 2021 toward more concentrated structural contributions linked to social and environmental conditions in 2023, offering a transparent methodological framework for comparative resilience analysis in small-sample policy settings.

One of the most well-known foundational concepts in decision analysis is the Ellsberg paradox. It shows that people prefer alternatives where the probabilities are objective compared to alternatives where the probabilities are vague (uncertain, ambiguous). This fact challenges the expected utility theory, which is founded on the idea that people base their decisions on probabilities and utilities of different outcomes. This article discusses this paradox and other similar decision analysis concepts and observations in relation to risk science and contemporary knowledge on how to conceptualize and characterize risk. The main aim of the article is to show that the Ellsberg paradox and related concepts and observations reinforce contemporary risk conceptualizations with their idea of seeing uncertainty as a main component of risk.