Probability-based Risk Research Articles

Multimodal learning system integrating electronic medical records and hysteroscopic images for reproductive outcome prediction and risk stratification of endometrial injury: a multicenter diagnostic study.

To develop a multimodal learning application system that integrates electronic medical records (EMR) and hysteroscopic images for reproductive outcome prediction and risk stratification of patients with intrauterine adhesions (IUAs) resulting from endometrial injuries. EMR and 5014 revisited hysteroscopic images of 753 post hysteroscopic adhesiolysis patients from the multicenter IUA database we established were randomly allocated to training, validation, and test datasets. The respective datasets were used for model development, tuning, and testing of the multimodal learning application. MobilenetV3 was employed for image feature extraction, and XGBoost for EMR and image feature ensemble learning. The performance of the application was compared against the single-modal approaches (EMR or hysteroscopic images), DeepSurv and ElasticNet models, along with the clinical scoring systems. The primary outcome was the 1-year conception prediction accuracy, and the secondary outcome was the assisted reproductive technology (ART) benefit ratio after risk stratification. The multimodal learning system exhibited superior performance in predicting conception within 1-year, achieving areas under the curves of 0.967 (95% CI: 0.950-0.985), 0.936 (95% CI: 0.883-0.989), and 0.965 (95% CI: 0.935-0.994) in the training, validation, and test datasets, respectively, surpassing single-modal approaches, other models and clinical scoring systems (all P<0.05). The application of the model operated seamlessly on the hysteroscopic platform, with an average analysis time of 3.7±0.8s per patient. By employing the application's conception probability-based risk stratification, mid-high-risk patients demonstrated a significant ART benefit (odds ratio=6, 95% CI: 1.27-27.8, P=0.02), while low-risk patients exhibited good natural conception potential, with no significant increase in conception rates from ART treatment (P=1). The multimodal learning system using hysteroscopic images and EMR demonstrates promise in accurately predicting the natural conception of patients with IUAs and providing effective postoperative stratification, potentially contributing to ART triage after IUA procedures.

AI-enhanced digital inspection of bridges

Civil infrastructure inspection -and consequently maintenance- is carried out primarily through visual inspections. AI-enhanced (Artificial Intelligence) digital inspection methods, integrated with risk-based probabilistic approaches, have been promoted to keep existing structures, especially infrastructures, safe and predictable. Drones are used to obtain a significant number of images to cover the surface of a bridge, which are further integrated into a digital 3D (three-dimensional) model. According to the IFC standards (Industry Foundation Class), this 3D model is GPS-positioned (Global Positioning System) and connected to BIM (Building Information Modelling). Post-processing the accumulated data volume of all digital images is very time-consuming. For this reason, appropriate AI-based algorithms streamline this process significantly, enabling partially automated damage detection and assessment. To this end, images of various types of damage on different bridges are used to train and test the AI-enhanced models. In addition, damage identification and classification are developed. Six visually detectable defects can be identified, and theoretical models estimate the associated structural diseases. Finally, a probability-based risk assessment presents the basis for defining the criticality of the structure. With the help of digital images, it is possible to create a high-fidelity digital model and quantitative surface and spatial data records of the structural health condition of bridges and other infrastructures.

Evaluation of future flood probability in agricultural reservoir watersheds using an integrated flood simulation system

In agricultural reservoir watersheds, the presence of various agricultural hydraulic structures combined with the influence of climate change necessitates effective flood risk analysis. This study aims to develop a flood risk evaluation technique for agricultural basins by incorporating the low temporal resolution of climate change scenarios. The approach utilizes flood probability as a key component and incorporates the Multiplicative Random Cascade model as a probability-based rainfall time decomposition model to capture the inherent uncertainty in climate change data. Reliability analysis with the limit state equation is employed for probability-based risk evaluation. Given that agricultural watersheds typically consist of agricultural reservoirs and farmland, the study develops a hydraulic and hydrological linkage model to comprehensively calculate flood probability, simulating from the agricultural reservoir in the upper side to the farmland located at the bottom edge of the watershed. The flood risk assessment is applied to the Yedang Reservoir watershed in Korea, using the Shared Socio-economic Pathways climate change scenario. Results indicate that flood probability is the highest in river levees during future precipitation events, converging to nearly 100% under 24-hour duration rainfall, irrespective of the rain frequency. Despite the hydraulic structures at the farmland being designed for a lower frequency of rain events, the average probability of flooding in the farmland have reached 50%. The agricultural reservoir exhibits a low risk of flooding even under climate change conditions. The proposed method not only enables the establishment of flood countermeasures but also facilitates the evaluation and prediction of future flood risk for each hydraulic structure located in rural areas affected by climate change.

A theory for combinations of risk measures

We study combinations of risk measures under no restrictive assumption on the set of alternatives. We develop and discuss results regarding the preservation of properties and acceptance sets for the combinations of risk measures. One of the main results is the representation for resulting risk measures from the properties of both alternative functionals and combination functions. To that, we build on the development of a representation for arbitrary mixture of convex risk measures. In this case, we obtain a penalty that recalls the notion of inf-convolution under theoretical measure integration. As an application, we address the context of probability-based risk measurements for functionals on the set of distribution functions. We develop results related to this specific context. We also explore features of individual interest generated by our framework, such as the preservation of continuity properties, the representation of worst-case risk measures, stochastic dominance and elicitability. We also address model uncertainty measurement under our framework and propose a new class of measures for this task.

Leveraging a Wildfire Risk Prediction Metric with Spatial Clustering

Fire authorities have started widely using operational fire simulations for effective wildfire management. The aggregation of the simulation outputs on a massive scale creates an opportunity to apply the evolving data-driven approach to closely estimate wildfire risks even without running computationally expensive simulations. In one of our previous works, we validated the application with a probability-based risk metric that gives a series of probability values for a fire starting at a start location under a given weather condition. The probability values indicate how likely it is that a fire will fall into different risk categories. The metric considered each fire start location as a unique entity. Such a provision in the metric could expose the metric to scalability issues when the metric is used for a larger geographic area and consequently make the metric hugely intensive to compute. In this work, in an investigative effort, we investigate whether the spatial clustering of fire start locations based on historical fire areas can address the issue without significantly compromising the accuracy of the metric. Our results show that spatially clustering all fire start locations in Tasmania into three risk clusters could leverage the probability-based risk metric by reducing the computational requirements of the metric by a theoretical factor in thousands with a mere compromise of approximately 5% in accuracy for two risk categories of high and low, thereby validating the possibility of the leverage of the metric with spatial clustering.

Assessment of Ship-Overtaking Situation Based on Swarm Intelligence Improved KDE.

This paper proposes a data-driven risk assessment model for ship overtaking based on the particle swarm optimization (PSO) improved kernel density estimation (KDE). By minimizing the mean square error between the real probability distribution of the ship overtaking point and the kernel density estimation probability distribution calculated by the current kernel density bandwidth, the longitude and latitude of the ship overtaking point are displayed by the color corresponding to the probability as the cost objective function of the search bandwidth of the algorithm. This can better show the distribution of the overtaking points of channel propagation traffic flow. A probability-based ship-overtaking risk evaluation model is developed through the bandwidth and density analysis optimized by an intelligent algorithm. In order to speed up searching the optimal variable width of the kernel density estimator for ship encountering positions, an improved adaptive variable-width kernel density estimator is proposed. The latter reduces the risk of too smooth probability density estimation phenomenon. Its convergence is proved. Finally, the model can efficiently evaluate the risk status of ship overtaking and provide navigational auxiliary decision support for pilots.

A probability-based risk metric for operational wildfire risk management

With the advancement in scientific understanding and computing technologies, fire practitioners have started relying on operational fire simulation tools to make better-informed decisions during wildfire emergencies. This increased use has created an opportunity to employ an emerging data-driven approach for wildfire risk estimation as an alternative to running computationally expensive simulations. In an investigative attempt, we propose a probability-based risk metric that gives a series of probability values for fires starting at any possible start location under any given weather condition falling into different categories. We investigate the validity of the proposed approach by applying it to use cases in Tasmania, Australia. Results show that the proposed risk metric can be a convenient and accurate method of estimating imminent risk during operational wildfire management. Additionally, the knowledge base of our proposed risk metric based on a data-driven approach can be constantly updated to improve its accuracy.

A probability-based risk assessment of roof strata weighting in longwall panels

Weighting of the roof strata in a longwall panel poses serious stability concerns at the working face. Therefore, it is essential to estimate weighting in terms of its interval. Furthermore, difficulty arises during the prediction of roof weighting as rock mass contains inherent uncertainties. In order to cater to this need, probabilistic analysis has been employed. The article aims to decipher the suitability of Monte Carlo simulation technique for the assessment of caving of roof strata. In the current study, firstly, regression models have been developed to estimate periodic weighting interval and first significant main roof weighting interval using data from thirty (30) longwall panels in India. The regression models were then used as deterministic mathematical models for Monte Carlo simulation. It has been observed that the Monte Carlo simulation technique is an acceptable means to evaluate the uncertainties arising from varying geo-mining conditions. In addition to it, sensitivity analysis results exhibited that the tensile strength and unit weight of the main roof is the key factor that governs the interval and severity of the periodic weighting and the first significant main roof weighting.

Generic algorithm to assess moisture susceptibility of simplified wall assemblies

A simplified wall assembly consisting of four material elements is considered. Optional vapor retarder and thermal resistances are located between the elements. A simplified numerical model is presented to perform probability-based risk analysis using the Monte Carlo method. The risk assessment of the wall assembly considers the interior and exterior climatic loads. The moisture transport mechanism accounted for in the analysis is diffusion. A prescribed rain load is assumed to be transferred directly to a chosen outer susceptible material element in the construction. A case study with a wooden wall structure is analyzed and the probability distribution of the time of wetness is presented for a case using 30 different climatic years, randomly varying indoor moisture excess, indoor temperature, and rain leakages. The results show that, in the case investigated in this paper, the simulations converged after 600 years. The proposed simplified model, combined with the use of Monte Carlo simulations, offers better opportunities for a systematic assessment of risk compared with the currently used simulation programs using only a design year.

The prediction of suicide in severe mental illness: development and validation of a clinical prediction rule (OxMIS)

Assessment of suicide risk in individuals with severe mental illness is currently inconsistent, and based on clinical decision-making with or without tools developed for other purposes. We aimed to develop and validate a predictive model for suicide using data from linked population-based registers in individuals with severe mental illness. A national cohort of 75,158 Swedish individuals aged 15–65 with a diagnosis of severe mental illness (schizophrenia-spectrum disorders, and bipolar disorder) with 574,018 clinical patient episodes between 2001 and 2008, split into development (58,771 patients, 494 suicides) and external validation (16,387 patients, 139 suicides) samples. A multivariable derivation model was developed to determine the strength of pre-specified routinely collected socio-demographic and clinical risk factors, and then tested in external validation. We measured discrimination and calibration for prediction of suicide at 1 year using specified risk cut-offs. A 17-item clinical risk prediction model for suicide was developed and showed moderately good measures of discrimination (c-index 0.71) and calibration. For risk of suicide at 1 year, using a pre-specified 1% cut-off, sensitivity was 55% (95% confidence interval [CI] 47–63%) and specificity was 75% (95% CI 74–75%). Positive and negative predictive values were 2% and 99%, respectively. The model was used to generate a simple freely available web-based probability-based risk calculator (Oxford Mental Illness and Suicide tool or OxMIS) without categorical cut-offs. A scalable prediction score for suicide in individuals with severe mental illness is feasible. If validated in other samples and linked to effective interventions, using a probability score may assist clinical decision-making.

Assessing the natural and anthropogenic influences on basin-wide fish species richness

Theory predicts that the number of fish species increases with river size in natural free-flowing rivers, but the relationship is lost under intensive exploitation of water resources associated with dams and/or landscape developments. In this paper, we aim to identify orthomorphic issues that disrupt theoretical species patterns based on a multi-year, basin-wide assessment in the Danshuei River Watershed of Taiwan. We hypothesize that multiple human-induced modifications fragment habitat areas leading to decreases of local fish species richness. We integrally relate natural and anthropogenic influences on fish species richness by a multiple linear regression model that is driven by a combination of factors including river network structure controls, water quality alterations of habitat, and disruption of channel connectivity with major discontinuities in habitat caused by dams.We found that stream order is a major forcing factor representing natural influence on fish species richness. In addition to stream order, we identified dams, dissolved oxygen deficiency (DO), and excessive total phosphorus (TP) as major anthropogenic influences on the richness of fish species. Our results showed that anthropogenic influences were operating at various spatial scales that inherently regulate the physical, chemical, and biological condition of fish habitats. Moreover, our probability-based risk assessment revealed causes of species richness reduction and opportunities for mitigation. Risks of species richness reduction caused by dams were determined by the position of dams and the contribution of tributaries in the drainage network. Risks associated with TP and DO were higher in human-activity-intensified downstream reaches. Our methodology provides a structural framework for assessing changes in basin-wide fish species richness under the mixed natural and human-modified river network and habitat conditions. Based on our analysis results, we recommend that a focus on landscape and riverine habitats and maintaining long-term monitoring programs are crucial for effective watershed management and river conservation plans.

Auditors' Risk Assessments: The Effects of Elicitation Approach and Assertion Framing

ABSTRACT This experimental study replicates, using U.S. audit practionioners, Fukukawa and Mock (2011), who investigated the effects of probability-based versus belief-based risk assessments and positive versus negative assertions on auditor judgments. Most results are consistent with the prior study: (1) significant differences between probability-based and belief-based risk assessments are observed; (2) assessed risks are significantly higher and relatively more skeptical when negatively versus positively stated assertions are provided; and (3) when the belief-based assessments are transformed into probabilities using a method proposed by Cobb and Shenoy (2006) and compared with the probability-based risk assessments, the difference is not statistically significant. Some results do not replicate, particularly when the risk assessments after audit evidence is provided are examined. However, in general, the U.S. results are more consistent with expectations. Overall, this study corroborates the key results of Fukukawa and Mock (2011). Data Availability: Contact the first author for data for use by others interested in extending or replicating results.

Ontology-based scenario modeling and analysis for bank stress testing

The 2008 banking crisis demonstrated that there is a lack of effective methods for modeling and analyzing “exceptional but plausible” risk scenarios in bank stress testing. Existing stress testing practices mainly focus on modeling probability-based risk factors and events in banking systems using historical data. Rare (low probability) risk events that can cause financial crises in banking systems, such as the bankruptcy of Lehman Brothers, are largely ignored due to the lack of appropriate modeling and analysis methods. To address this problem, we propose an approach called Banking Event-driven Scenario-oriented Stress Testing (or simply, BESST) which has two main components: 1) an ontology-based event-driven scenario model (OESM), and 2) two analysis methods based on OESM for scenario recommendation and plausibility checking. The proposed BESST approach provides bank stress testing stakeholders an effective method for modeling and analyzing financial crisis scenarios that are rare but often have significant consequences.

A Probabilistic Landscape Analysis Supporting the Management of Unplanned Ignitions at Sequoia and Kings Canyon National Parks

Fire management policy in the United States has evolved to encourage fire managers to address the full range of beneficial and detrimental effects of unplanned ignitions. We developed a probability-based fire risk and benefit simulation approach to support wildland fire management at Sequoia and Kings Canyon National Parks (SEKI) in a manner consistent with the SEKI fire management plan. The approach integrated fire behavior, fire spread probability, and resource values to estimate the spatial distribution of expected fire net benefits for any ignition point in a landscape context. We simulated probabilistic fire footprints from each burnable cell and evaluated them for their potential to provide ecosystem benefits or to require risk mitigation. The expected net benefits of igniting a cell were stored in the ignition cell to provide a landscape-level perspective. We provide results that reflect the fire planning weather conditions of concern to demonstrate how the sensitivity of the results can reflect weather influences. The results provided a preprocessed “playbook” type reference guide to fire management. The results can be quickly accessed and intuitively interpreted and were used to promote interagency cooperation in managing the Tehipite fire on northern SEKI and on the Sierra National Forest.

Probability-based assessment of dam safety using combined risk analysis and reliability methods – application to hazards studies

Dam safety assessment is a major challenge for engineers specialising in hydraulic works. It is now standardised in regulations providing for hazards studies using probability-based risk analysis procedures. Dam safety probability-based assessment includes two key scientific issues: the modelling of complex mechanisms that occur within the works, and a probability-based assessment of structural safety. This paper introduces a combined application of risk analysis and dam reliability methodologies. This combination required the completion of our work on the adaptation of risk analysis methods to dams, and on the probability-based modelling of strength and loading input data for a reliability-based analysis. These methodologies are then implemented as part of a hazards study on a roller-compacted-concrete gravity dam.

Solution method of overtopping risk model for earth dams

Hydrologic risk analysis relies on a series of probabilistic analyses, and it is a complex problem in estimating the probability distributions of multiple independent and random variables. The goal of this study is to presents the procedure and application of a probability-based risk analysis methodology to evaluate earth dam overtopping risk that induced by concurrent flood and wind. The uncertainty arising from initial water surface level, flood, wind velocity, and dam height are discussed in this research. The improved Monte Carlo simulation and mean-value first-order second-moment method are used to solve the proposed dam overtopping risk model, respectively. The nonparametric kernel density estimation method, which can better learn the complex multimodal characteristic of probability density function than that of traditional parametric estimation method, is employed to improve the probability density function of initial water surface level. The latin hypercube sampling is introduced to generate uniform random number, which improves the efficient and stability compared with simple random sampling. Afterward, an application to the Dongwushi Reservoir in China illustrates that the dam overtopping risk computed using the improved Monte Carlo simulation is lower than that using mean-value first-order second-moment method. Furthermore, the sensitivity analysis show that initial water surface level is more sensitive to overtopping risk than wind velocity.

A risk concept applicable for both probabilistic and non-probabilistic perspectives

In engineering risk assessments, probability is the common tool used to describe the epistemic uncertainties about unknown quantities. Probability is considered a main component of risk. However, a number of alternative approaches exist for representing and describing uncertainties in risk assessments, including possibility theory and evidence theory. For these approaches and theories a probability-based risk definition cannot in general serve as a conceptual framework for risk assessments. A broader risk perspective is required. The purpose of the present paper is to present such a perspective and show how both the probabilistic and the alternative approaches and theories can be supported by this perspective. The key feature of this perspective is that uncertainty replaces probability in the definition of risk.

Audit Risk Assessments Using Belief versus Probability

SUMMARY: Important audit judgments concerning client selection, audit program planning, and risk assessment all may be impacted by the approach taken to assess risk and to interpret audit evidence. This experimental study examines whether auditors’ risk assessments are influenced by the risk assessment approach and by assertion framing. Risk assessment approach was manipulated by eliciting belief-based versus probability-based risk assessments. Assertion framing was manipulated by stating the financial statement assertions to be examined in a positive versus negative form.Four measures of risk of material misstatement are compared, including one based on the Cobb and Shenoy (2003a, 2003b, 2006) transformation of beliefs into “probabilities.” Both risk assessment approach and assertion framing effects on the auditors’ risk assessments are observed. Given these significant differences, the risk measure(s) the auditor chooses to focus on and the way assertions are framed are shown to be important audit choices in that such choices may impact both audit effectiveness and efficiency.

Auditors’ evidence evaluation and aggregation using beliefs and probabilities

This paper examines possible differences in auditors’ performance when they make belief-based versus probability-based risk assessments by focusing on two phases of the financial statement audit process: the assessment of two attributes of audit evidence (‘strength’ and ‘direction’) and the aggregation of evidence. Based on an experiment in which 48 experienced auditors participated, three important findings were observed. First, there was no significant difference in the mean assessment of strength of evidence measured using the likelihood ratio. However, the difference in the assessed direction of evidence, that is whether the evidence is interpreted as being confirming or disconfirming, is significant for one of the cases examined. This result shows that auditors making belief-based assessments are able to assess the direction of the evidence more accurately than auditors making probability-based assessments. Third, the auditors’ aggregation of evidence was not in accordance with ‘AND’ logic for either auditors making belief-based or probability-based assessments. These empirical results raise issues which need to be addressed in practice and in future research.

Excess risk thresholds in ultrasound safety studies: Statistical methods for data on occurrence and size of lesions

Concerns about the safe use of clinical ultrasound (US) at diagnostic pressure levels (below a mechanical index, or MI, = 1.9) have stimulated considerable research in US risk assessment. The objective of the present study was to develop probability-based risk thresholds for US safety studies, to present statistical methods for estimating the thresholds and their standard errors and to compare these methods with the analysis based on a piecewise linear (“hockey stick”) model. The excess risk at exposure level x > 0 was defined as the relative increase in the probability of a lesion at that level compared with the background probability of a lesion at exposure x = 0. The risk threshold was then defined as the exposure level at which the excess risk exceeded a specified level (e.g. 5% or 50%). Thus, given pressure-dependent estimates of the excess risk, the thresholds were estimated by solving the risk equation to obtain the pressure at which the target level of excess risk occurs. Threshold estimates of this type have been developed extensively in the literature for incidence (presence or absence) data. Only recently, however, have excess risk threshold estimates been derived for data in which lesion size (depth, surface area) is measured if present and a zero is recorded if the lesion is absent. Tobit regression was used to estimate pressure-dependent percentiles of the size distribution, and the excess risks were estimated from the tobit probability of a positive-valued response. The tobit model provides a well-established approach to modeling data constrained to be nonnegative. Solving the risk equation for the tobit model leads to risk threshold estimates that incorporate the information on size of observed lesions. Results using these probability-based risk estimates were compared with results for a piecewise linear (“hockey stick”) model, which has also been used in the US safety literature, although it does not explicitly address the nonnegativity constraint in the sampling model. The comparisons were carried out for data from two previously published studies, from different laboratories, on US-induced lung hemorrhage. The thresholds derived from logistic regression of lesion occurrence and tobit regression of lesion size were quite consistent with each other and within sampling error. The hockey stick thresholds, defined as the exposure level at which the piecewise linear model for the probability of the expected size of a lesion bends upward, corresponded to quite different excess risk values for incidence (lesion occurrence) compared with size (lesion surface area or depth), although these methods have been developed previously for both types of data. The use of probability-based excess risk thresholds is recommended to obtain consistent incidence vs. size thresholds and to ensure that the thresholds are well-defined and interpretable independent of the details of the statistical model. (E-mail: dgs@uiuc.edu)