Bayesian Belief Network Model Research Articles

Time-series forecasting of road distress parameters using dynamic Bayesian belief networks

PurposeTo proactively draw efficient maintenance plans, road agencies should be able to forecast main road distress parameters, such as cracking, rutting, deflection and International Roughness Index (IRI). Nonetheless, the behavior of those parameters throughout pavement life cycles is associated with high uncertainty, resulting from various interrelated factors that fluctuate over time. This study aims to propose the use of dynamic Bayesian belief networks for the development of time-series prediction models to probabilistically forecast road distress parameters.Design/methodology/approachWhile Bayesian belief network (BBN) has the merit of capturing uncertainty associated with variables in a domain, dynamic BBNs, in particular, are deemed ideal for forecasting road distress over time due to its Markovian and invariant transition probability properties. Four dynamic BBN models are developed to represent rutting, deflection, cracking and IRI, using pavement data collected from 32 major road sections in the United Arab Emirates between 2013 and 2019. Those models are based on several factors affecting pavement deterioration, which are classified into three categories traffic factors, environmental factors and road-specific factors.FindingsThe four developed performance prediction models achieved an overall precision and reliability rate of over 80%.Originality/valueThe proposed approach provides flexibility to illustrate road conditions under various scenarios, which is beneficial for pavement maintainers in obtaining a realistic representation of expected future road conditions, where maintenance efforts could be prioritized and optimized.

Bayesian belief network modelling approach for predicting and ranking risk factors for malaria infections among children under 5 years in refugee settlements in Uganda

BackgroundMalaria risk factors at household level are known to be complex, uncertain, stochastic, nonlinear, and multidimensional. The interplay among these factors, makes targeted interventions, and resource allocation for malaria control challenging. However, few studies have demonstrated malaria’s transmission complexity, control, and integrated modelling, with no available evidence on Uganda’s refugee settlements. Using the 2018–2019 Uganda’s Malaria Indicator Survey (UMIS) data, an alternative Bayesian belief network (BBN) modelling approach was used to analyse, predict, rank and illustrate the conceptual reasoning, and complex causal relationships among the risk factors for malaria infections among children under-five in refugee settlements of Uganda.MethodsIn the UMIS, household level information was obtained using standardized questionnaires, and a total of 675 children under 5 years were tested for malaria. From the dataset, a casefile containing malaria test results, demographic, social-economic and environmental information was created. The casefile was divided into a training (80%, n = 540) and testing (20%, n = 135) datasets. The training dataset was used to develop the BBN model following well established guidelines. The testing dataset was used to evaluate model performance.ResultsModel accuracy was 91.11% with an area under the receiver-operating characteristic curve of 0.95. The model’s spherical payoff was 0.91, with the logarithmic, and quadratic losses of 0.36, and 0.16 respectively, indicating a strong predictive, and classification ability of the model. The probability of refugee children testing positive, and negative for malaria was 48.1% and 51.9% respectively. The top ranked malaria risk factors based on the sensitivity analysis included: (1) age of child; (2) roof materials (i.e., thatch roofs); (3) wall materials (i.e., poles with mud and thatch walls); (4) whether children sleep under insecticide-treated nets; 5) type of toilet facility used (i.e., no toilet facility, and pit latrines with slabs); (6) walk time distance to water sources (between 0 and 10 min); (7) drinking water sources (i.e., open water sources, and piped water on premises).ConclusionRanking, rather than the statistical significance of the malaria risk factors, is crucial as an approach to applied research, as it helps stakeholders determine how to allocate resources for targeted malaria interventions within the constraints of limited funding in the refugee settlements.

Exploring Global Competitiveness Index 4.0 through the lens of country risk

The Global Competitiveness Index (GCI) 4.0 provides an annual evaluation of country-level productivity and development in the fourth industrial revolution era. The GCI 4.0 is classified into 12 pillars representing socioeconomic, financial, health, technological, and other factors of production. Country risk, which is an aggregate measure of country-level exposure to financial, political, and economic risks, can influence national development. This study explores the association between country risk and competitiveness by mapping country risk drivers to GCI and GCI pillars to country risk. Two Bayesian Belief Network models, with a minimum prediction accuracy of 81.3 %, are developed using two data-sets representing the GCI and country risk ratings of 107 countries. Sensitivity analysis reveals that low performance in information and communication technology (ICT) adoption (monetary policy) can significantly affect country risk (GCI), whereas innovation capability (financial system risk) and institutions (legal system) can considerably improve country risk (GCI). By delving into how the pillars of competitiveness interact with various risk factors such as political stability, economic volatility, and regulatory environment, decision-makers, investors, and policymakers can gain valuable insights that guide effective resource allocation, risk mitigation strategies, informed policy formulation, and sustainable development efforts.

Accommodating uncertainty in soil erosion risk assessment: Integration of Bayesian belief networks and MPSIAC model

Accommodating uncertainty stands as one of the most salient challenges in the development of soil erosion assessment tools. We presented a novel approach integrating the Modified Pacific Southwest Inter-Agency Committee (MPSIAC) model and Bayesian Belief Networks (BBNs) to assess soil erosion in a region of western Iran. The soil erosion status was reckoned based on the nine factors of MPSIAC. We utilized BBNs to produce a causal model for soil erosion, with output probabilities being validated through re-evaluation and sensitivity analysis. We identified erosion types, geological formations, run-off, soil erodibility, soil permeability, soil characteristics, and precipitation intensity as the main determinants of soil erosion. A significant, positive correlation existed between the erosion rate derived from MPSIAC and BBNs model in all land-use/covers over the work units. Overall, this study highlighted the potential of BBNs as a supportive tool for soil erosion prediction as well as a relatively simple and updatable soil erosion model for dealing with the diagnostic, scenario, and sensitivity analysis. Considering the increasing incidence of soil erosion, the BBNs model proposed in this study can be extended to a variety of ecosystems that are subject to soil erosion and changes in the probability of its causal factors.

Exploring network-based dependencies between country-level sustainability and business risks

The concept of sustainability has gained prominence in the global discourse, with increasing recognition of its impact on business risk. By integrating sustainability considerations into their strategies and operations, national policymakers and businesses can mitigate business risks and contribute to national sustainability goals. The objective of this study is to explore dependencies among the risks associated with the 17 Sustainable Development Goals (SDGs), introduced by the United Nations in 2015, and country-level business risk. In particular, this study aims to identify critical SDGs that can influence business risk based on the two extreme performance states (i.e., best and worst scenarios) of individual SDGs. A data-driven Bayesian Belief Network model is developed using two datasets on country-level sustainability performance and business risk assessment. On the one hand, ‘quality education’, ‘no poverty’, ‘affordable and clean energy’ and ‘sustainable cities and communities’ are identified as the most critical SDGs based on the negative impact of their associated risks on business risk. On the other hand, low risk (high performance) associated with the ‘peace, justice and strong institutions’, ‘life on land’ and ‘industry, innovation and infrastructure’ SDGs can reduce business risk. Strong dependencies are found between individual SDG risks and business risk, implying policymakers need to consider the implications of their sustainability initiatives on the risks and uncertainties surrounding their country's business environment.

Bayesian belief network model for quantifying the delay risk of marine construction projects in developing countries

Bayesian belief network model for quantifying the delay risk of marine construction projects in developing countries

A data-driven Bayesian belief network model for exploring patient experience drivers in healthcare sector

A data-driven Bayesian belief network model for exploring patient experience drivers in healthcare sector

Failure assessment of oil and gas transmission pipelines using an integrated Bayesian belief network and GIS model

Corrosion is one of the main reasons for pipeline failure in the oil and gas industry. Because a pipeline failure can result in serious personal injury, monetary loss, and environmental damage, pipeline operators need to make timely, and cost-effective decisions to prevent accidents in high consequence areas. The majority of models proposed in this area are designed and computed for individual pipe segments, with only a limited number of studies considering the system level. The system level pertains to spatially distributed pipelines that span hundreds of miles across different environmental conditions. The current study propose combining GIS and Bayesian belief network to determine the probability of failure of transmission pipelines due to external pitting corrosion. The model incorporates data from an extensive network of pipes spanning hundreds of miles and their surroundings to compute the failure probability of the pipeline infrastructure. The combination of spatial GIS capabilities with the reasoning capabilities of the Bayesian network provides a powerful tool to estimate the likelihood of transmission pipelines failing in a specific area, based on the available information.

A geochemically informed leak detection (GILD) model for CO2 injection sites

Early detection of CO2 leakage through monitoring is important to ensure long-term safety for geologic carbon storage (GCS). A geochemically informed leak detection (GILD) model has been developed for groundwater chemistry monitoring at CO2 injection sites. The GILD model integrates a geochemical model that simulates fluid chemistry changes in CO2 leakage events and a Bayesian belief network (BBN) model that evaluates monitoring observations to identify leakages. The geochemical model is implemented using Geochemists’ Workbench to assess fluid chemistry changes as a result of small CO2 leakage in an above-zone monitoring interval (AZMI) formation with varying mineral assemblages and background fluids. Response functions are fitted to the output of the geochemical model and are translated to conditional probabilities in the BBN model. The BBN model gives operational prediction of the leak probability given a set of groundwater monitoring measurements and the probability of detecting a leak at a given magnitude. The detection capabilities of multiple monitoring parameters are compared. For aquifers that contain calcite, it is valuable to incorporate other monitoring parameters with pH to increase the sensitivity of detection. For aquifers with no calcite, pH alone is a sensitive parameter. This research illustrates a method of identifying CO2 leakage into aquifers with both geochemical and statistical tools.

Analysis of Carbon Emissions and Influencing Factors in China Based on City Scale

Assessing regional carbon emissions and their relationship with socio-economic conditions is very important for developing strategies for carbon emission reduction. This study explored the impact of the proportion of non-fossil energy, the land development degree, the urbanization rate of permanent residents, the proportion of secondary industry, per capita GDP, and per capita construction land area on per capita CO2 emissions in 339 prefecture-level and above cities in China (excluding some cities in Xinjiang, Hong Kong, Macao, and Taiwan). A Bayesian belief network modeling carbon emissions was constructed to identify the global effects of various factors on per capita CO2 emissions, and multiscale geographically weighted regression was used to analyze their local effects. The results showed that first, per capita CO2 emissions of cities in China increased from the south to the north and decreased from the eastern coast to the inland region. Second, globally, the sensitivity of per capita CO2 emissions to various factors from high to low was in the order of per capita construction land area>per capita GDP>urbanization rate of permanent residents>land development degree>proportion of secondary industry>proportion of non-fossil energy. Third, locally, the direction of the spatial relationship between each factor and per capita CO2 emissions was consistent with the global relationship, and there was spatial heterogeneity in the strength of the relationship. Finally, clean energy, decarbonization technologies, saving and intensive use of land, and green living were effective ways to achieve the dual-carbon goal.

Modelling the Influential Factors Embedded in the Proportionality Assessment in Military Operations

The ongoing decade was believed to be a peaceful one. However, contemporary conflicts, and in particular, ongoing wars prove the opposite as they show the increase in context complexity when defining their goals as well as execution strategies for building means and methods for achieving them by gaining advantage against their adversaries through the engagement of well-established targets. At the core of the engagement decision relies the principle of proportionality which brings in a direct relation the expected unintended effects on civilian side with the anticipated intended effects on military side. While the clusters of effects involved in the proportionality assessment are clear, the process itself is subjective, governed by different dimensions of uncertainty, and represents the responsibility of military Commanders. Thus, a complex socio-technical process where different clusters of influential factors (e.g., military, technical, socio-ethical) play a role in the decisions made. Having said that, the objective of this research is to capture and cluster these factors, and further to model their influence in the proportionality decision-making process. This decision support system produces military targeting awareness to the agents involved in the processes of building, executing, and assessing military operations. To accomplish the aim of this research, a Design Science Research methodological approach is taken for capturing and modelling the influential factors as a socio-technical artefact in the form of a Bayesian Belief Network (BBN) model. The model proposed is further evaluated through demonstration on three different cases in respect to real military operations incidents and scenarios existing in the scientific literature in this research field. Hence, through this demonstration, it is illustrated and interpreted how the factors identified influence proportionality decisions when assessing target engagement as being proportional or disproportional. In these cases, corresponding measures for strengthening proportionality and reducing disproportionality in military operations are considered.

Environmental Risk Assessment of Wetland Ecosystems Using Bayesian Belief Networks

Wetlands are valuable natural capital and sensitive ecosystems facing significant risks from anthropogenic and climatic stressors. An assessment of the environmental risk levels for wetlands’ dynamic ecosystems can provide a better understanding of their current ecosystem health and functions. Different levels of environmental risk are defined by considering the categories of risk and the probability and severity of each in the environment. Determining environmental risk levels provides a general overview of ecosystem function. This mechanism increases the visibility of risk levels and their values in three distinct states (i.e., low, moderate, and high) associated with ecosystem function. The Bayesian belief network (BBN) is a novel tool for determining environmental risk levels and monitoring the effectiveness of environmental planning and management measures in reducing the levels of risk. This study develops a robust methodological framework for determining the overall level of risks based on a combination of varied environmental risk factors using the BBN model. The proposed model is adopted for a case study of Shadegan International Wetlands (SIWs), which consist of a series of Ramsar wetlands in the southwest of Iran with international ecological significance. A comprehensive list of parameters and variables contributing to the environmental risk for the wetlands and their relationships were identified through a review of literature and expert judgment to develop an influence diagram. The BBN model is adopted for the case study location by determining the states of variables in the network and filling the probability distribution tables. The environmental risk levels for the SIWs are determined based on the results obtained at the output node of the BBN. A sensitivity analysis is performed for the BBN model. We proposed model-informed management strategies for wetland risk control. According to the BBN model results, the SIWs ecosystems are under threat from a high level of environmental risk. Prolonged drought has been identified as the primary contributor to the SIWs’ environmental risk levels.

Pathways for improving rangeland governance under constraining land tenure systems: Application of a participatory Bayesian Belief approach.

This paper analyzes the complex relationships of factors influencing rangeland governance in the arid areas. A Bayesian Belief Network (BBN) model was developed in a participatory way to illustrate and assess the weight of a combination of environmental, social and institutional factors enabling the achievement of strengthened rangeland governance at a local level. The Bayesian model was applied in Tataouine situated in south of Tunisia. A complex diagram was built in a participatory way to illustrate the most important interactions between rangeland socioecological system components. Then, a BBN model was applied for a predictive purpose, by inserting evidence conditional probabilities on the most frequent land tenure systems in the region, and thus exploring pathways to improve rangeland governance under each of these systems. Results show that overall improvement of rangeland governance in the study area is highly related to the type of tenure system, performances of farmers organizations, clarification of boundaries between neighboring rangelands, and diversification of the production systems in place by including other agricultural activities in addition to pastoralism. Pastoral areas with tenure systems characterized by combined grazing of private and collective rangelands were revealed to be the most constraining for successful rangeland governance. However, even under such tenure system, the improvement of a set of institutional attributes, especially in relation to the empowerment of farmers organizations can help improving local governance.

Bayesian Belief Network Modeling of Accident Occurrence in Metal Lathe Machining Operations

Accidents occurrence in metal lathe machining operations in industrial workshops often cost organizations billions of dollars while injured workers and families are faced with financial and emotional burdens. Studies revealed that the fly-out accident is the most probable accident that occurs during metal lathe machining operations. The uncertainty surrounding its occurrence is rarely reported. This study, therefore, modeled the uncertainty surrounding the occurrence of a fly-out accident during metal lathe machining operations and its corresponding consequences using the Bayesian belief network (BBN). Fly-out accident causal factors were identified representing the parent nodes with two states each. Two child-node scenarios were modeled on Bayesian belief influence diagrams, namely the fly-out accident with two states (yes and no) and the consequences of the fly-out accident with three states (fatal, serious and minor). Seven causal factors of the fly-out accident were identified (chuck-related fault, tool-post failure, workpiece holding fault, coolant fault, wrong operating speed, safety-related guards fault and wrong feed rate). Bayesian causal inference of fly-out accident was 0.708 and the fatal fly-out accident was 0.263. Bayesian diagnostic inference showed that chuck association fault and improper feed rate were significant causal factors influencing the occurrence of a fly-out accident, fatal fly-out accident and serious fly-out accident, while the occurrence of a minor fly-out accident was affected by coolant fault during machining operations. The study identified areas of safety concerns that may be used for the development of Machine Workshop Safety Management Systems toward sustainable, safe, and effective machine workshop operations.

Construction Dispute Potentials: Mechanism versus Empiricism in Artificial Neural Networks

The booming development of neural network algorithms has shifted the research focus in the field of construction project management from causal investigation to statistical approximation and hence from mechanistic models to empirical models. This paper took construction dispute avoidance as an example and enabled the best efforts to establish paired mechanistic and empirical models to investigate if the pursuit of a mechanistic understanding of construction disputes should be continued. A Bayesian belief network and multilayer perceptron were used for mechanistic and empirical simulations, respectively. A list of critical dispute factors was first identified from the literature and shortlisted by Pearson’s chi-square tests and Pearson product-moment correlational coefficient tests. The structure of the Bayesian belief network was constructed with logical deduction assisted by a further literature review and Delphi surveys. A structured questionnaire survey was conducted to collect quantitative data for factor shortlisting and model quantification. It was revealed that, being assisted with machine learning techniques, both mechanistic and empirical models achieved an accuracy rate of over 95% under ideal conditions. However, Bayesian belief network models predicted better with fewer constraints due to their advantages in reflecting the formation mechanism of construction disputes, while multilayer perceptron models were more constrained by the inconvenience of sourcing high-quality data as model input. This paper demonstrated that it is still necessary to investigate the formation mechanism of construction disputes further for more efficient avoidance strategies. During the investigation of model construction and comparison, this paper also reflected on the interpretation of statistical threshold and proposed that an arbitrary single cut-off point for statistical tests could potentially eliminate factors that should have been included.

Effect of human–elephant conflict on local attitudes toward the conservation of wild Asian elephants in Myanmar

ABSTRACT The study of attitudes toward wild elephants and human–elephant conflict (HEC) is vital to understanding what attitudes are held by local people and how to incorporate them into wild elephant conservation. This study investigated the interlinkages between the HEC experience and local people’s attitudes toward the conservation of wild elephants and which exploratory factors influence these attitudes. We used a Bayesian Belief Network (BBN) framework to highlight the interlinkages between HEC and people’s attitudes. The basic BBN model indicated that HEC was central in determining the attitudes of local people. Although people generally hold positive attitudes toward elephants, people support having elephants in the country but not in their own regions. Conservation willingness was not obvious due to the lack of deliberate assistance from the government to the affected communities. We suggest implementing education programs to promote local awareness of conflict mitigation techniques.

Bayesian belief network-based human reliability analysis methodology for start-up and shutdown operations in nuclear power plants

Human-induced unplanned reactor trips during start-up and shutdown operations can impair the safety condition of the plant or threaten important plant safety functions. Despite the need for efforts to reduce such human errors through human reliability analysis (HRA), most existing HRA focuses on emergency or full-power operations. To address this, the current study proposes an HRA methodology for start-up and shutdown operations in which a quantitative evaluation is performed using a Bayesian belief network (BBN) model based on a developed quantification framework for the start-up and shutdown operation tasks. Before the evaluation, general operating procedure (GOP) analysis, simulation-based screening analysis, and task analysis are performed. To demonstrate the process of proposed methodology and its effectiveness, a case study with a GOP is conducted. Then to verify the feasibility of the proposed methodology, two comparison cases are conducted with BBN model evaluation and actual events that occurred in Korean NPPs.

Using weighted expert judgement and nonlinear data analysis to improve Bayesian belief network models for riverine ecosystem services

Rivers are a key part of the hydrological cycle and a vital conduit of water resources, but are under increasing threat from anthropogenic pressures. Linking pressures with ecosystem services is challenging because the processes interconnecting the physico-chemical, biological and socio-economic elements are usually captured using heterogenous methods. Our objectives were, firstly, to advance an existing proof-of-principle Bayesian belief network (BBN) model for integration of ecosystem services considerations into river management. We causally linked catchment stressors with ecosystem services using weighted evidence from an expert workshop (capturing confidence among expert groups), legislation and published literature. The BBN was calibrated with analyses of national monitoring data (including non-linear relationships and ecologically meaningful breakpoints) and expert judgement. We used a novel expected index of desirability to quantify the model outputs. Secondly, we applied the BBN to three case study catchments in Ireland to demonstrate the implications of changes in stressor levels for ecosystem services in different settings. Four out of the seven significant relationships in data analyses were non-linear, highlighting that non-linearity is common in ecosystems, but rarely considered in environmental modelling. Deficiency of riparian shading was identified as a prevalent and strong influence, which should be addressed to improve a broad range of societal benefits, particularly in the catchments where riparian shading is scarce. Sediment load had a lower influence on river biology in flashy rivers where it has less potential to settle out. Sediment interacted synergistically with organic matter and phosphate where these stressors were active; tackling these stressor pairs simultaneously can yield additional societal benefits compared to the sum of their individual influences, which highlights the value of integrated management. Our BBN model can be parametrised for other Irish catchments whereas elements of our approach, including the expected index of desirability, can be adapted globally.

Modelling services provisioning through tree species in the Moribane Forest Reserve, Mozambique

Among the main effects of human activities on the environment are land use and the resulting land cover changes. Such changes impact the capacity of ecosystems to provide goods and services to human beings. The benefits obtained by people from ecosystems are called ecosystem services. This paper aims at developing a model of ecosystem services provisioning through trees in the Moribane Forest Reserve (MFR), Manica - Mozambique. The study was carried out using a social survey through semi-structured interviews to 149 out of 479 households living within the reserve, and its objective was to understand the level of supply and demand for wood products. A field survey through 72 rectangular 50mx20m plots was also conducted with 24 plots established in each of the three land types, namely Forest land Fallow land and Cropland. This was to understand the phytosociological aspects of the most used wood species for household livelihood. The results show that, of the 89 identified trees, 61 are used locally for different purposes, including house building, manufacture of utensils and fuelwood. The Fallow land is the main source of extraction of such resources. The Forests revealed higher levels of richness in terms of species (83.1%). Thirty-five point nine (35.9%) of these species are assumed to be the most used by the community. Analysis through the Bayesian Belief Networks model shows a trend towards unsustainability, where over 80% of landscape units (plots) in Fallow and Cropland are of Moderate and Low Importance, evidencing the risk of unavailability of ecosystem services in the near future if the use and management practices are not changed. Conclusion, the model shows that the stage of services provided by wood forest species in MFR is under threat due to swidden-fallow rotations agriculture practice combined with overpressure. The study can help the academic community in future studies by replicating the methodology used for monitoring purposes or conducting studies in other similar areas, and the results may support decision-makers in designing better strategies towards sustainability. • The forest ecosystems constitute habitat and, a key provider of goods and services. • People in rural areas of developing countries are very reliant on tree provisioning services for their livelihoods. • The agriculture and exploitation of natural resources create serious concern when ecosystem changes affect ecological processes. • The alfa diversity and the structure are important metrics for monitoring ecological conditions. • The Bayesian Belief Networks are useful tools for representing expert knowledge about an ecosystem.

Weighted Bayesian Belief Network: A Computational Intelligence Approach for Predictive Modeling in Clinical Datasets.

There are growing concerns about the mortality due to Breast cancer many of which often result from delayed detection and treatment. So an effective computational approach is needed to develop a predictive model which will help patients and physicians to manage the situation timely. This study presented a Weighted Bayesian Belief Network (WBBN) modeling for breast cancer prediction using the UCI breast cancer dataset. New automated ranking method was used to assign proper weights to attribute value pair based on their impact on causing the disease. Association between attributes was generated using weighted association rule mining between two attributes, multiattributes, and with class labels to generate rules. Weighted Bayesian confidence and weighted Bayesian lift measures were used to produce strong rules to build the model. To build WBBN, the Open Markov tool was used for structure and parametric learning using generated strong rules. The model was trained using 70% records and tested on 30% records with a threshold value of minimum support = 36% and confidence = 70% which produced results with an accuracy of 97.18%. Experimental results show that WBBN achieved better results in most cases compared to other predictive models. The study would contribute to the fight against breast cancer and the quality of treatment.