"This Is What We Don't Know": Treating Epistemic Uncertainty in Bayesian Networks for Risk Assessment

Causal, predictive or observational? Different understandings of key event relationships for adverse outcome pathways and their implications on practice

The Adverse Outcome Pathways (AOPs) framework is pivotal in toxicology, but the, terminology describing Key Event Relationships (KERs) varies within AOP guidelines.This study examined the usage of causal, observational and predictive terms in AOP, documentation and their adaptation in AOP development. A literature search and text, analysis of key AOP guidance documents revealed nuanced usage of these terms, with KERs often described as both causal and predictive. The adaptation of, terminology varies across AOP development stages. Evaluation of KER causality often, relies targeted blocking experiments and weight-of-evidence assessments in the, putative and qualitative stages. Our findings highlight a potential mismatch between,terminology in guidelines and methodologies in practice, particularly in inferring,causality from predictive models. We argue for careful consideration of terms like, causal and essential to facilitate interdisciplinary communication. Furthermore, integrating known causality into quantitative AOP models remains a challenge.

Advancing exposure assessment approaches to improve wildlife risk assessment

The exposure assessment component of a Wildlife Ecological Risk Assessment aims to estimate the magnitude, frequency, and duration of exposure to a chemical or environmental contaminant, along with characteristics of the exposed population. This can be challenging in wildlife as there is often high uncertainty and error caused by broad-based, interspecific extrapolation and assumptions often because of a lack of data. Both the US Environmental Protection Agency (USEPA) and European Food Safety Authority (EFSA) have broadly directed exposure assessments to include estimates of the quantity (dose or concentration), frequency, and duration of exposure to a contaminant of interest while considering "all relevant factors." This ambiguity in the inclusion or exclusion of specific factors (e.g., individual and species-specific biology, diet, or proportion time in treated or contaminated area) can significantly influence the overall risk characterization. In this review, we identify four discrete categories of complexity that should be considered in an exposure assessment-chemical, environmental, organismal, and ecological. These may require more data, but a degree of inclusion at all stages of the risk assessment is critical to moving beyond screening-level methods that have a high degree of uncertainty and suffer from conservatism and a lack of realism. We demonstrate that there are many existing and emerging scientific tools and cross-cutting solutions for tackling exposure complexity. To foster greater application of these methods in wildlife exposure assessments, we present a new framework for risk assessors to construct an "exposure matrix." Using three case studies, we illustrate how the matrix can better inform, integrate, and more transparently communicate the important elements of complexity and realism in exposure assessments for wildlife. Modernizing wildlife exposure assessments is long overdue and will require improved collaboration, data sharing, application of standardized exposure scenarios, better communication of assumptions and uncertainty, and postregulatory tracking. Integr Environ Assess Manag 2024;20:674-698. © 2023 SETAC.

Causal diagramming for assessing human system risk in spaceflight

For over a decade, the National Aeronautics and Space Administration (NASA) has tracked and configuration-managed approximately 30 risks that affect astronaut health and performance before, during and after spaceflight. The Human System Risk Board (HSRB) at NASA Johnson Space Center is responsible for setting the official risk posture for each of the human system risks and determining-based on evaluation of the available evidence-when that risk posture changes. The ultimate purpose of tracking and researching these risks is to find ways to reduce spaceflight-induced risk to astronauts. The adverse effects of spaceflight begin at launch and continue throughout the duration of the mission, and in some cases, across the lifetime of the astronaut. Historically, research has been conducted in individual risk "silos" to characterize risk, however, astronauts are exposed to all risks simultaneously. In January of 2020, the HSRB at NASA began assessing the potential value of causal diagramming as a tool to facilitate understanding of the complex causes and effects that contribute to spaceflight-induced human system risk. Causal diagrams in the form of directed acyclic graphs (DAGs) are used to provide HSRB stakeholders with a shared mental model of the causal flow of risk. While primarily improving communication among those stakeholders, DAGs also allow a composite risk network to be created that can be tracked and configuration managed. This paper outlines the HSRB's pilot process for this effort, the lessons learned, and future goals for data-driven risk management approaches.

Hierarchical Analysis Process for Belief Management in Internet of Drones

Group awareness is playing a major role in the efficiency of mission planning and decision-making processes, particularly those involving spatially distributed collaborative entities. The performance of this concept has remarkably increased with the advent of the Internet of Things (IoT). Indeed, a myriad of innovative devices are being extensively deployed to collaboratively recognize and track events, objects, and activities of interest. A wide range of IoT-based approaches have focused on representing and managing shared information through formal operators for group awareness. However, despite their proven results, these approaches are still refrained by the inaccuracy of information being shared between the collaborating distributed entities. In order to address this issue, we propose in this paper a new belief-management-based model for a collaborative Internet of Drones (IoD). The proposed model allows drones to decide the most appropriate operators to apply in order to manage the uncertainty of perceived or received information in different situations. This model uses Hierarchical Analysis Process (AHP) with Subjective Logic (SL) to represent and combine opinions of different sources. We focus on purely collaborative drone networks where the group awareness will also be provided as service to collaborating entities.

Bayesian Network Applications for Sustainable Holistic Water Resources Management: Modeling Opportunities for South Africa

Anthropogenic transformation of land globally is threatening water resources in terms of quality and availability. Managing water resources to ensure sustainable utilization is important for a semiarid country such as South Africa. Bayesian networks (BNs) are probabilistic graphical models that have been applied globally to a range of water resources management studies; however, there has been very limited application of BNs to similar studies in South Africa. This article explores the benefits and challenges of BN application in the context of water resources management, specifically in relation to South Africa. A brief overview describes BNs, followed by details of some of the possible opportunities for BNs to benefit water resources management. These include the ability to use quantitative and qualitative information, data, and expert knowledge. BN models can be integrated into geographic information systems and predict impact of ecosystem services and sustainability indicators. With additional data and information, BNs can be updated, allowing for integration into an adaptive management process. Challenges in the application of BNs include oversimplification of complex systems, constraints of BNs with categorical nodes for continuous variables, unclear use of expert knowledge, and treatment of uncertainty. BNs have tremendous potential to guide decision making by providing a holistic approach to water resources management.

Causal Approach to Determining the Environmental Risks of Seabed Mining

Mineral deposits containing commercially exploitable metals are of interest for seabed mineral extraction in both the deep sea and shallow sea areas. However, the development of seafloor mining is underpinned by high uncertainties on the implementation of the activities and their consequences for the environment. To avoid unbridled expansion of maritime activities, the environmental risks of new types of activities should be carefully evaluated prior to permitting them, yet observational data on the impacts is mostly missing. Here, we examine the environmental risks of seabed mining using a causal, probabilistic network approach. Drawing on a series of expert interviews, we outline the cause-effect pathways related to seabed mining activities to inform quantitative risk assessments. The approach consists of (1) iterative model building with experts to identify the causal connections between seabed mining activities and the affected ecosystem components and (2) quantitative probabilistic modeling. We demonstrate the approach in the Baltic Sea, where seabed mining been has tested and the ecosystem is well studied. The model is used to provide estimates of mortality of benthic fauna under alternative mining scenarios, offering a quantitative means to highlight the uncertainties around the impacts of mining. We further outline requirements for operationalizing quantitative risk assessments in data-poor cases, highlighting the importance of a predictive approach to risk identification. The model can be used to support permitting processes by providing a more comprehensive description of the potential environmental impacts of seabed resource use, allowing iterative updating of the model as new information becomes available.

Increased Use of Bayesian Network Models Has Improved Environmental Risk Assessments

Environmental and ecological risk assessments are defined as the process for evaluating the likelihood that the environment may be impacted as a result of exposure to stressors. Although this definition implies the calculation of probabilities, risk assessments traditionally rely on nonprobabilistic methods such as calculation of a risk quotient. Bayesian network (BN) models are a tool for probabilistic and causal modeling, increasingly used in many fields of environmental science. Bayesian networks are defined as directed acyclic graphs where the causal relationships and the associated uncertainty are quantified in conditional probability tables. Bayesian networks inherently incorporate uncertainty and can integrate a variety of information types, including expert elicitation. During the last 2 decades, there has been a steady increase in reports on BN applications in environmental risk assessment and management. At recent annual meetings of the Society of Environmental Toxicology and Chemistry (SETAC) North America and SETAC Europe, a number of applications of BN models were presented along with new theoretical developments. Likewise, recent meetings of the European Geosciences Union (EGU) have dedicated sessions to Bayesian modeling in relation to water quality. This special series contains 10 articles based on presentations in these sessions, reflecting a range of BN applications to systems, ranging from cells and populations to watersheds and national scale. The articles report on recent progress in many topics, including climate and management scenarios, ecological and socioeconomic endpoints, machine learning, diagnostic inference, and model evaluation. They demonstrate that BNs can be adapted to established conceptual frameworks used to support environmental risk assessments, such as adverse outcome pathways and the relative risk model. The contributions from EGU demonstrate recent advancements in areas such as spatial (geographic information system [GIS]-based) and temporal (dynamic) BN modeling. In conclusion, this special series supports the prediction that increased use of Bayesian network models will improve environmental risk assessments. Integr Environ Assess Manag 2021;17:53-61. © 2020 The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC).

Quantification of an Adverse Outcome Pathway Network by Bayesian Regression and Bayesian Network Modeling

The adverse outcome pathway (AOP) framework has gained international recognition as a systematic approach linking mechanistic processes to toxicity endpoints. Nevertheless, successful implementation into risk assessments is still limited by the lack of quantitative AOP models (qAOPs) and assessment of uncertainties. The few published qAOP models so far are typically based on data-demanding systems biology models. Here, we propose a less data-demanding approach for quantification of AOPs and AOP networks, based on regression modeling and Bayesian networks (BNs). We demonstrate this approach with the proposed AOP #245, "Uncoupling of photophosphorylation leading to reduced ATP production associated growth inhibition," using a small experimental data set from exposure of Lemna minor to the pesticide 3,5-dichlorophenol. The AOP-BN reflects the network structure of AOP #245 containing 2 molecular initiating events (MIEs), 3 key events (KEs), and 1 adverse outcome (AO). First, for each dose-response and response-response (KE) relationship, we quantify the causal relationship by Bayesian regression modeling. The regression models correspond to dose-response functions commonly applied in ecotoxicology. Secondly, we apply the fitted regression models with associated uncertainty to simulate 10 000 response values along the predictor gradient. Thirdly, we use the simulated values to parameterize the conditional probability tables of the BN model. The quantified AOP-BN model can be run in several directions: 1) prognostic inference, run forward from the stressor node to predict the AO level; 2) diagnostic inference, run backward from the AO node; and 3) omnidirectionally, run from the intermediate MIEs and/or KEs. Internal validation shows that the AOP-BN can obtain a high accuracy rate, when run is from intermediate nodes and when a low resolution is acceptable for the AO. Although the performance of this AOP-BN is limited by the small data set, our study demonstrates a proof-of-concept: the combined use of Bayesian regression modeling and Bayesian network modeling for quantifying AOPs. Integr Environ Assess Manag 2021;17:147-164. © 2020 The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC).