Metamodeling Approach Research Articles

Recalibrating an Established Microsimulation Model to Capture Trends and Projections of Colorectal Cancer Incidence and Mortality.

Changing colorectal cancer (CRC) incidence rates, including recent increases for people younger than 50 y, need to be considered in planning for future cancer control and screening initiatives. Reliable estimates of the impact of changing CRC trends on the National Bowel Cancer Screening Program (NBCSP) are essential for programmatic planning in Australia. An existing microsimulation model of CRC, Policy1-Bowel, was updated to reproduce Australian CRC trends data and provide updated projections of CRC- and screening-related outcomes to inform clinical practice guidelines for the prevention of CRC. Policy1-Bowel was recalibrated to reproduce statistical age-period-cohort model trends and projections of CRC incidence for 1995-2045 in the absence of the NBCSP as well as published data on CRC incidence trends, stage distribution, and survival in 1995-2020 in Australia. The recalibrated Policy1-Bowel predictions were validated by comparison with published Australian CRC mortality trends for 1995-2015 and statistical projections to 2040. Metamodels were developed to aid the calibration process and significantly reduce the computational burden. Policy1-Bowel was recalibrated, and best-fit parameter sets were identified for lesion incidence, CRC stage progression rates, detection rates, and survival rates by age, sex, bowel location, cancer stage, and birth year. The recalibrated model was validated and successfully reproduced observed CRC mortality rates for 1995-2015 and statistical projections for 2016-2030. The recalibrated Policy1-Bowel model captures significant additional detail on the future incidence and mortality burden of CRC in Australia. This is particularly relevant as younger cohorts with higher CRC incidence rates approach screening ages to inform decision making for these groups. The metamodeling approach allows fast recalibration and makes regular updates to incorporate new evidence feasible. In Australia, colorectal cancer incidence rates are increasing for people younger than 50 y but decreasing for people older than 50 y, and colorectal cancer survival is improving as new treatment technologies emerge.To evaluate the future health and economic impact of screening and inform policy, modeling must include detailed trends and projections of colorectal cancer incidence, mortality, and diagnosis stage.We used novel techniques including integrative age-period cohort projections and metamodel calibration to update Policy1-Bowel, a detailed microsimulation of colorectal cancer and screening in Australia.

A Metamodeling Approach for Planning Critical IoT Systems

This paper presents a metamodeling approach to address the lack of methodological support in canvas model development, focusing on planning critical IoT systems. We introduce MM4Canvas - a metamodel that provides a solid foundation for developing structured and standardized canvas models, allowing for consistent reuse and extension across diverse project types. A proof of concept was conducted by instantiating a general-purpose canvas model based on MM4Canvas for project planning, aiming to establish a connection between this activity and the Requirements Engineering process. We extended this model to incorporate safety and security requirements for critical IoT systems. Our contribution demonstrates the metamodel’s capacity to support standardization, reuse, and extensibility in canvas-based project planning.

Synthesizing Local Capacities, Multi-Source Remote Sensing and Meta-Learning to Optimize Forest Carbon Assessment in Data-Poor Regions

As the climate emergency escalates, the role of forests in carbon sequestration is paramount. This paper proposes a framework that integrates local capacities, multi-source remote sensing data, and meta-learning to enhance forest carbon assessment methodologies in data-scarce regions. By integrating multi-source optical and radar remote sensing data alongside community forest inventories, we applied a meta-modelling approach using stacked generalization ensemble to estimate forest above-ground carbon (AGC). We also conducted a Kruskal–Wallis test to determine significant differences in AGC among different tree species. The Kruskal–Wallis test (p = 1.37 × 10−13) and Dunn post-hoc analysis revealed significant differences in carbon stock potential among tree species, with Afzelia quanzensis (x~ = 12 kg/ha, P-holm-adj. = 0.05) and the locally known species M’buta (x~ = 6 kg/ha, P-holm-adj. = 5.45 × 10−9) exhibiting a significantly higher median AGC. Our results further showed that combining optical and radar remote sensing data substantially improved prediction accuracy compared to single-source remote sensing data. To improve forest carbon assessment, we employed stacked generalization, combining multiple machine learning algorithms to leverage their complementary strengths and address individual limitations. This ensemble approach yielded more robust estimates than conventional methods. Notably, a stacking ensemble of support vector machines and random forest achieved the highest accuracy (R2 = 0.84, RMSE = 1.36), followed by an ensemble of all base learners (R2 = 0.83, RMSE = 1.39). Additionally, our results demonstrate that factors such as the diversity of base learners and the sensitivity of meta-leaners to optimization can influence stacking performance.

Mapping the spatial transferability of knowledge-guided machine learning: Application to the prediction of drain flow fraction.

Machine learning (ML) methods continue to gain traction in hydrological sciences for predicting variables at large scales. Yet, the spatial transferability of these ML methods remains a critical yet underexamined aspect. We present a metamodel approach to obtain large-scale estimates of drain fraction at 10m spatial resolution, using a ML algorithm (Gradient Boost Decision Tree). Our variable of interest is drain, as artificial drainage of agricultural land is widespread in areas with high groundwater tables. Drainage has significant effects on the hydrological cycle, and impacts groundwater recharge, streamflow partitioning and nutrient transport. Drain flow is controlled by small-scale variations in topography, geology and groundwater depth, which presents challenges to its estimation at large scale. Drain fraction is the average ratio between drain flow and precipitation. The metamodel combines covariates based on topography, land use and geology with simulated drain fraction from 45 field-scalephysically-based hydrological models of Danish drain catchments. The 45 models were jointly calibrated against timeseries of drain flow observations. The metamodel was used to upscale predictions of drain fractions for the entirety of Danish agricultural land. This involved considerable extrapolation beyond the 45 drain catchments used for training, calling for an assessment of spatial transferability. To map transferability of the model, and distinguish areas where metamodel results are reliable or not, we used the concept of area of applicability (AOA). The AOA is determined from the similarity of covariate space covered by the training data compared to each prediction point, assuming a correlation between model performance and covariate similarity. AOA mapping showed 71% of Denmark's agricultural land falling within the AOA of the metamodel. The study presents a stepwise methodology to obtain national-scale results using a ML model trained on local-scale numeric models and an evaluation of its spatial transferability, highly relevant for decision-support purposes.

Improved Prediction of Ligand-Protein Binding Affinities by Meta-modeling.

The accurate screening of candidate drug ligands against target proteins through computational approaches is of prime interest to drug development efforts. Such virtual screening depends in part on methods to predict the binding affinity between ligands and proteins. Many computational models for binding affinity prediction have been developed, but with varying results across targets. Given that ensembling or meta-modeling approaches have shown great promise in reducing model-specific biases, we develop a framework to integrate published force-field-based empirical docking and sequence-based deep learning models. In building this framework, we evaluate many combinations of individual base models, training databases, and several meta-modeling approaches. We show that many of our meta-models significantly improve affinity predictions over base models. Our best meta-models achieve comparable performance to state-of-the-art deep learning tools exclusively based on 3D structures while allowing for improved database scalability and flexibility through the explicit inclusion of features such as physicochemical properties or molecular descriptors. We further demonstrate improved generalization capability by our models using a large-scale benchmark of affinity prediction as well as a virtual screening application benchmark. Overall, we demonstrate that diverse modeling approaches can be ensembled together to gain meaningful improvement in binding affinity prediction.

Numerical evaluation of wear parameters using meta-models

Wear parameters identified by linear friction tester (LFT) experiments overestimate the wear mass loss when applied on a laboratory abrasion & skid tester 100 (LAT100) simulation. On the other hand, the identification of wear parameters directly from the LAT100 experiment can be challenging since variables such as the contact area and the slip velocity are not measured experimentally and need to be assumed. To improve the identification process, numerical calibration is used as a more suitable approach. This approach relies on meta-models as a target function for minimization. The meta-models are generated using a sample of wear parameters applied to an LAT100 finite element modeling (FEM) to calculate the corresponding wear mass.. In this model, a transport velocity is defined for the rolling simulation, and an arbitrary lagrangian–eulerian (ALE) adaptive meshing approach is adopted for the wear modeling. For the wear model, a combination of archard’s wear model and schallamach’s abrasion law is used. This model contains a pair of wear parameters to be identified. The ALE adaptive meshing technique moves the nodes independently of the material. Since the mesh topology remains the same, failure of the simulation occurs if the wear volume loss exceeds that of the element. Meta-models are created to extend wear modeling beyond this failure. Once the meta-models are created, they are used as a target function for the minimization algorithm. The minimization algorithm aims to find the optimal wear parameters by minimizing the difference between experimentally observed and numerically produced wear mass loss. The minimization algorithm inputs a set of wear parameters into the meta-models which in turn yield a prediction of the wear mass loss. The process is carried out until an optimum parameter set is identified. Such an approach has a lower accuracy if the parameters are identified directly from the experiment using assumptions regarding the contact shear stress and the sliding velocity. Nonetheless, the main advantage of parameters identified using the meta-model approach is the usability of these parameters in an LAT100 model.

A physics-informed neural network enhanced importance sampling (PINN-IS) for data-free reliability analysis

Reliability analysis of highly sensitive structures is crucial to prevent catastrophic failures and ensure safety. Therefore, these safety-critical systems are to be designed for extremely rare failure events. Accurate statistical quantification of these events associated with a very low probability of occurrence requires millions of evaluations of the limit state function (LSF) involving computationally expensive numerical simulations. Variance reduction techniques like importance sampling (IS) reduce such repetitions to a few thousand. The use of a data-centric metamodel can further cut it down to a few hundred. In data-centric metamodeling approaches, the actual complex numerical analysis is performed at a few points to train the metamodel for approximating the structural response. On the other hand, a physics-informed neural network (PINN) can predict the structural response based on the governing differential equation describing the physics of the problem, without a single evaluation of the complex numerical solver, i.e., data-free. However, the existing PINN models for reliability analysis have been effective only in estimating a large range of failure probabilities (10-1∼10-3). To address this issue, the present study develops a PINN-based data-free reliability analysis for low failure probabilities (<10-5). In doing so, a two-stage PINN integrated with IS (PINN-IS) is proposed. The first stage is employed to approximate the most probable failure point (MPP) appropriately while the second stage enhances the accuracy of LSF predictions at the IS population centred on the approximated MPP. The effectiveness of the proposed approach is numerically illustrated by three structural reliability analysis examples.

Corn yield response to phosphorus fertilizer in Michigan: a metamodeling approach for phosphorus management policies

Phosphorus (P) pollution from nonpoint sources is a persistent agricultural externality. P fertilizer use in excess of crop removal leads to soil P accumulation and increases chance of P runoff from agricultural fields. For fertilizer-intensive crops such as corn, effective nutrient management can partly mitigate the externality associated with fertilizer use. In states such as Michigan with abundant freshwater resources that are vulnerable to P pollution and where corn is grown intensively, understanding yield response is instrumental for P management policies. We analyze corn yield response to P fertilizer and control for soil P. We use farmer survey data from 1650 corn farmers in Michigan and a crop-simulation model to obtain data in a metamodeling approach to estimate yield function. Using the estimated function in profit maximization set up, we calculate agronomically and economically optimal profit-maximizing P application rates for Michigan, across agricultural regions, and across different soil types. We find that corn yield responds to P only in low soil P (0–5 ppm or mg kg−1) fields and the economically optimal P rate for Michigan is [63, 68] kg ha−1 or [53, 61] lbs acre−1 for low soil P fields and would be zero for fields with adequate soil P. For existing soil P levels in Michigan, the self-reported P application of Michigan corn farmers exceeds the profit-maximizing P application rate. We compare our analysis results with self-reported survey data to demonstrate excess P applications. Since many fields in Michigan have high soil P, additional P application in excess of crop removal increases risks of eutrophication. We also find that 10 % increase in P prices faced by farmers would lead to 3.1 % reduction in the economically optimal P application rate. Based on these results, we discuss suitable policy options such as fertilizer “usage fee” and tailored fertilizer recommendations.

IAT/ML: a metamodel and modelling approach for discourse analysis

Language technologies are gaining momentum as textual information saturates social networks and media outlets, compounded by the growing role of fake news and disinformation. In this context, approaches to represent and analyse public speeches, news releases, social media posts and other types of discourses are becoming crucial. Although there is a large body of literature on text-based machine learning, it tends to focus on lexical and syntactical issues rather than semantic or pragmatic. Being useful, these advances cannot tackle the nuanced and highly context-dependent problems of discourse evaluation that society demands. In this paper, we present IAT/ML, a metamodel and modelling approach to represent and analyse discourses. IAT/ML focuses on semantic and pragmatic issues, thus tackling a little researched area in language technologies. It does so by combining three different modelling approaches: ontological, which focuses on what the discourse is about; argumentation, which deals with how the text justifies what it says; and agency, which provides insights into the speakers’ beliefs, desires and intentions. Together, these three modelling approaches make IAT/ML a comprehensive solution to represent and analyse complex discourses towards their understanding, evaluation and fact checking.

Artificial neural network infused quasi oppositional learning partial reinforcement algorithm for structural design optimization of vehicle suspension components

Abstract This paper introduces and investigates an enhanced Partial Reinforcement Optimization Algorithm (E-PROA), a novel evolutionary algorithm inspired by partial reinforcement theory to efficiently solve complex engineering optimization problems. The proposed algorithm combines the Partial Reinforcement Optimization Algorithm (PROA) with a quasi-oppositional learning approach to improve the performance of the pure PROA. The E-PROA was applied to five distinct engineering design components: speed reducer design, step-cone pulley weight optimization, economic optimization of cantilever beams, coupling with bolted rim optimization, and vehicle suspension arm optimization problems. An artificial neural network as a metamodeling approach is used to obtain equations for shape optimization. Comparative analyses with other benchmark algorithms, such as the ship rescue optimization algorithm, mountain gazelle optimizer, and cheetah optimization algorithm, demonstrated the superior performance of E-PROA in terms of convergence rate, solution quality, and computational efficiency. The results indicate that E-PROA holds excellent promise as a technique for addressing complex engineering optimization problems.

Managing low–acuity patients in an Emergency Department through simulation–based multiobjective optimization using a neural network metamodel

This paper deals with Emergency Department (ED) fast-tracks for low-acuity patients, a strategy often adopted to reduce ED overcrowding. We focus on optimizing resource allocation in minor injuries units, which are the ED units that can treat low-acuity patients, with the aim of minimizing patient waiting times and ED operating costs. We formulate this problem as a general multiobjective simulation-based optimization problem where some of the objectives are expensive black-box functions that can only be evaluated through a time-consuming simulation. To efficiently solve this problem, we propose a metamodeling approach that uses an artificial neural network to replace a black-box objective function with a suitable model. This approach allows us to obtain a set of Pareto optimal points for the multiobjective problem we consider, from which decision-makers can select the most appropriate solutions for different situations. We present the results of computational experiments conducted on a real case study involving the ED of a large hospital in Italy. The results show the reliability and effectiveness of our proposed approach, compared to the standard approach based on derivative-free optimization.

A reliability-based design against post-buckling load drop in spherical shell cap with stochastic imperfections

Elastic buckling is a critical design consideration for deep spherical shell caps. Significant reduction is observed in the experimental critical load from the linearized eigen-buckling analysis, for reasons extensively discussed in the past and relooked recently through the energy barrier concepts. The Knock Down Factors (KDFs) were proposed to account for such a reduction. The highly conservative KDFs constitute the lower bound of a large experimental dataset; that can be reconciled by nonlinear stability analysis of the imperfect shell. This study models the KDFs as Uncertain-but-Bounded (UBB) variables to present a Reliability-Based Design (RBD) for spherical shell caps having random geometric imperfections in radius (R) and thickness (t). The RBD furnishes the pertinent design variables ((R/t) ratio and the dome height-to-base radius (H/r0) ratio) for a target reliability index (β) and the respective KDFs. A metamodelling approach is employed to reduce the computational burden of the proposed RBD for a thin spherical shell cap having zero-mean, stationary, homogeneous, gaussian stochastic geometric imperfections. The Riks algorithm is employed in the commercial code ABAQUS for the finite element-based nonlinear stability analysis of shells. The KDFs and the reliability bounds are presented for varying degrees of imperfections. Remarkably, the margin of conservatism in the KDFs is shown to be narrowed down by incorporating the dependency on (H/r0).

IoT Security Model for Smart Cities based on a Metamodeling Approach

IoT Security Model for Smart Cities based on a Metamodeling Approach

Efficient MAPoD via Least Angle Regression based Polynomial Chaos Expansion Metamodel for Eddy Current NDT

In this article, a metamodeling approach based on non-intrusive polynomial chaos expansion (PCE) with least angle regression (LAR) method is used in boundary element analysis for a model-assisted probability of detection (MAPoD) study of eddy current nondestructive testing (NDT) systems. The LAR-PCE metamodel represents the NDT system model responses by a set of coefficients with the polynomial basis functions in lieu of pure kernel degeneration accelerated boundary element method (KD-BEM) based physical model. Both the computational accuracy and efficiency of the LAR-PCE metamodel over the ordinary least squares (OLS) based PCE metamodel are demonstrated by testing the 3D eddy current NDT benchmarks with different system setups, flaw lengths and widths. The simulation results show two digits accuracy of the PoD metrics compared with the ones achieved by the KD-BEM based physical model as the benchmark. The LAR-PCE metamodel has remarkable improvements in computational efficiency over the OLS-PCE metamodel and accelerates the MAPoD study.

Design and validation of a new ring hoop plane strain test for characterizing the anisotropic plastic behavior of tubular materials

The accurate characterization of the plastic behavior of tubular materials is challenging due to the difficulties in acquiring proper specimens for mechanical testing. In this study, we propose a novel testing technique, namely the ring hoop plane strain Test (RHPST), as a supplementary test for characterizing materials experiencing distinct strain paths compared to standard uniaxial tensile tests. To design a suitable ring specimen’s geometry that satisfies the necessary plane strain conditions, we developed a meta-modelling approach merging response surface method (RSM) and the finite element method (FEM). Experimental tests are conducted on RHPST specimens, and 3D digital image correlation (DIC) technique is used to monitor the strain fields within the specimen's gauge section. However, observations indicate the presence of edge effects on the gauge specimen. These edge effects significantly influence and limit the homogeneity of the plane strain field. To tackle this limitation, a correction procedure is developed for separating the authentic plane strain area, constituting approximately 78% of the RHPST specimen's width. Furthermore, we successfully fabricated a novel grooved ring specimen (GRHPST), which exhibits a wider region of the plane strain state encompassing up to 90% of the gauge width, while minimizing the impact of edge effects. The true stress-true strain curve obtained from the GRHPST specimen exhibits exceptional agreement with the curve predicted using the Hill48 yield criterion. Notably, no additional corrections are required for the measured true stress-true strain curve, thereby affirming the efficacy of the GRHPST specimen as an adept test for characterizing the anisotropic plastic behavior of AA6063 tubes.

Consistent meta-modelling approach for a less-conservative reliability based design of shells against buckling

Consistent meta-modelling approach for a less-conservative reliability based design of shells against buckling

Optimization of a Tapered Specimen Geometry for Short-Term Dynamic Tensile Testing of Continuous Fiber Reinforced Thermoplastics

Continuous fiber reinforced thermoplastics (cFRTP) are one of the most promising lightweight materials. For their use in structural components, reproducible and comparable material values have to be evaluated, especially at high strain rates. Due to their high stiffness and outstanding strength properties, the evaluation of the material behavior at high strain rates is complex. In the presented work, a new tensile specimen geometry for strain rate testing is virtually optimized using a metamodel approach with an artificial neural network. The final specimen design is experimentally validated and compared with rectangular specimen results for a carbon fiber reinforced polycarbonate (CF-PC). The optimized specimen geometry leads to 100% valid test results in experimental validation of cross-ply laminates and reaches 9% higher tensile strength values than the rectangle geometry with applied end tabs at a strain rate of 40 s−1. Through the optimization, comparable material parameters can be efficiently generated for a successful cFRTP strain rate characterization.

Metamodeling on uncertainty quantification in the behavior of the tire/road interaction of vehicles

In recent years, the automotive industry has been developing applied research to meet customer’s needs; considering safety, vehicle comfort and energetic efficiency. In particular, automotive tires have a prominent position in this research area, ensuring good vehicle handling, comfort and safety. In the vehicle dynamics performance, excellent gripping and reduced rolling resistance in the tires are crucial to maximize the energetic efficiency in a reliable way. However, to ensure such a level of reliability in vehicle operation, the inherent uncertainties of tires, as well as other factors subject to variability must be taken into account in the vehicle design. In this way, the present paper analyzes in detail the effect of variations in some parameters such as ambient temperature, ground conditions, vertical load, speed and tire inflation on the analysis of vehicle dynamics. A Metamodeling approach associated with the Monte Carlo Simulation was employed to develop the mathematical models to analyze the effect of uncertain parameters on the tire rolling resistance; traction, centripetal and lateral forces, using experimental data from the literature, in the longitudinal and lateral vehicle dynamics. Therefore, the present research brings as an innovation an integrated approach to the input parameters of the system with the rolling resistance through the developed metamodels. There was a substantial variability of up to 15% both up and down in the Maximum Traction Force of a vehicle in response to variations in the vehicle’s weight and the coefficient of tire rolling resistance. In contrast, the Lateral Force exhibited a greater variability, with a 25 downward and 10% upward variation associated with the weight and friction coefficient variability of the vehicle. Further investigations into the sensitivity analysis highlight the significant influence of the friction coefficient and temperature on the Traction Forces of the vehicle.

A Shrinkage Approach to Improve Direct Bootstrap Resampling Under Input Uncertainty

Discrete-event simulation models generate random variates from input distributions and compute outputs according to the simulation logic. The input distributions are typically fitted to finite real-world data and thus are subject to estimation errors that can propagate to the simulation outputs: an issue commonly known as input uncertainty (IU). This paper investigates quantifying IU using the output confidence intervals (CIs) computed from bootstrap quantile estimators. The standard direct bootstrap method has overcoverage due to convolution of the simulation error and IU; however, the brute-force way of washing away the former is computationally demanding. We present two new bootstrap methods to enhance direct resampling in both statistical and computational efficiencies using shrinkage strategies to down-scale the variabilities encapsulated in the CIs. Our asymptotic analysis shows how both approaches produce tight CIs accounting for IU under limited input data and simulation effort along with the simulation sample-size requirements relative to the input data size. We demonstrate performances of the shrinkage strategies with several numerical experiments and investigate the conditions under which each method performs well. We also show advantages of nonparametric approaches over parametric bootstrap when the distribution family is misspecified and over metamodel approaches when the dimension of the distribution parameters is high. History: Accepted by Bruno Tuffin, Area Editor for Simulation. Funding: This work was supported by the National Science Foundation [CAREER CMMI-1834710, CAREER CMMI-2045400, DMS-1854659, and IIS-1849280]. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2022.0044 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2022.0044 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .

Structuring and organizing database security domain from big data perspective using meta-modeling approach

Database security is an area focused on safeguarding databases against harmful access. It involves ensuring data accuracy, blocking unauthorized entry, and preventing harmful code within the database. Although various security models and methods exist, they often don't comprehensively cover all aspects of database security. This leads to a diverse and unclear understanding of database security among experts. This study proposes a unified framework, the Database Security Meta-model (DBSM), which acts as a standard language in this field. The DBSM, comprising twelve main elements, is thoroughly vetted to align with security needs and offers guidelines for practitioners to create specific security solutions.