Heterogeneous Model Research Articles

Preliminary Study on Multi-Scale Modeling of Asphalt Materials: Evaluation of Material Behavior through an RVE-Based Approach

This study employs a microstructure-based finite element modeling approach to understand the mechanical behavior of asphalt mixtures across different length scales. Specifically, this work aims to develop a multi-scale modeling approach employing representative volume elements (RVEs) of optimal size; this is a key issue in asphalt modeling for high-fidelity fracture modeling of heterogeneous asphalt mixtures. To determine the optimal RVE size, a convergence analysis of homogenized elastic properties is conducted using two types of RVEs, one made with polydisperse spherical inclusions, and another made with polydisperse truncated cylindrical inclusions, each aligned with the American Association of State Highway and Transportation Official’s maximum density gradation curve for a 12.5 mm Nominal Maximum Aggregate Size (NMAS). The minimum RVE lengths for this NMAS were found to be in the range of 32–34 mm. After the optimal RVE size for each inclusion shape is obtained, computational models of heterogeneous Indirect Tensile Asphalt Cracking Test samples are then generated. These models include the components of viscoelastic mastic, linear elastic aggregates, and cohesive zone modeling to simulate the rate-dependent failure evolution from micro- to macro-cracking. Examination of load-displacement responses at multiple loading rates shows that both heterogeneous models replicate experimentally measured data satisfactorily. Through micro- and macro-level analyses, this study enhances our understanding of the composition-performance relationships in asphalt pavement materials. The procedure proposed in this study allows us to identify the optimal RVE sizes that preserve computational efficiency without significantly compromising their ability to capture the asphalt material behavior under specific operational conditions.

Digital Twin Reveals the Impact of Carbon Binder Domain Distribution on Performance of Lithium‐Ion Battery Cathodes

The rising demand for high‐performance lithium‐ion batteries (LIBs) emphasizes the need for precise electrode design. The carbon binder domain (CBD) within electrodes, crucial for electron transport and structural integrity, can impede lithium‐ion transport and reduce electrochemically active sites. This study leverages digital twin technology to investigate the impact of CBD distribution along the electrode thickness on LIB cathode performance. Virtual LIB cathodes with varying CBD configurations are generated, and their performance using 3D heterogeneous electrochemical modeling is evaluated. The results reveal that a steep CBD gradient significantly diminishes charge capacity compared to a mild gradient, particularly at higher current rates. This reduction is attributed to an increased tortuosity factor and an uneven distribution of electrochemical activity, which adversely affect mass transport and electrochemical reaction dynamics. These effects are more pronounced at lower porosities and higher loading levels. These findings highlight that optimizing CBD distribution is critical for advancing electrode design and enhancing LIB cathode performance.

Efficacy and safety of short segment pedicle screw fixation in patients with neurologically intact burst fractures of the lower thoracic and lumbar spine: a meta-analysis of studies published over the last 20 years

Objective. To conduct a systematic review and meta-analysis of studies on the surgical treatment of patients with uncomplicated burst fractures of the lower thoracic and lumbar spine and to determine the effectiveness and safety of short transpedicular fixation (TPF) in this patient group.Material and Methods. The study included articles with the following criteria: publication date from January 1, 2004, to December 31, 2023; patient sample descriptions involving uncomplicated burst fractures from T10 to L5; TPF involving one segment adjacent to the fractured vertebra in both cranial and caudal directions without spinal fusion; descriptions of treatment outcomes or complications; and an average follow-up period of at least 12 months. Meta-analysis was conducted using the Comprehensive Meta-Analysis software, version 2.2.064. Depending on the level of heterogeneity (I² test), either a fixed-effects or random-effects model was applied. Begg’s or Egger’s test was used to assess publication bias, and any bias present was corrected using the trim-and-fill method.Results. The application of TPF resulted in a significant reduction in the overall Cobb angle by 5.9 degrees in the percutaneous group and by 7.6 degrees when using a midline approach. Regarding AVBCR (anterior vertebral body compression ratio), a reduction of 24.0 % and 24.8 % was observed in both groups, respectively. The overall complication rates were as follows: superficial infection, 2.2 %; deep infection, 2.0 %; and implant-associated complications, 5.6 %. No patient developed a neurological deficit. The levels of work adaptation W1 and W2 on the Denis scale were achieved in 70.9 % of patients. The overall quality of life, as measured by the Oswestry Disability Index, averaged 13.4 %.Conclusions. Short transpedicular fixation without additional spinal fusion or laminectomy appears to be an effective and safe method for treating burst fractures of the lower thoracic and lumbar spine without neurological deficits. This method allows for regression of kyphotic deformity in the long-term post-injury period by at least 5.9 degrees and restoration of anterior vertebral height by 24 %. The approach demonstrated relatively low overall postoperative complication rates. More than 90 % of patients were able to return to full-time work, either in their previous position or with reduced physical demands.

Heterogeneous temperature characteristics considering rainfall effect during construction of an integrally-poured rock-filled concrete arch dam

Integrally-poured arch dam with the material of rock-filled concrete (RFC) is a new and rapid construction technology for mass concrete structures. There is still limited literature regarding the heterogeneous temperature field of RFC, especially under special weather conditions. In this study, a novel mesoscale model of heterogeneous RFC was generated through a combination of the discrete element method (DEM), the finite element method (FEM), and the background grid mapping method (BGMM). Based on the monitoring experiments at the Longdongwan project, this study greatly enhances the simulation precision and details by incorporating a two-phase material of heterogeneous RFC model, i.e. the stochastic pre-laid rocks and layered SCC (self-compacting concrete), and refining the analysis time step to an hourly basis. The results show that: (1) An infinite heat convection coefficient on the exposed RFC surfaces was effective in simulating the rainfall effect, and the mesoscopic simulation results aligned well with the monitoring data. (2)The cold wave with rainfall can cause a notable drop of around 10 °C in the placement temperatures of both rocks and SCC, and lowered the temperature rise by about 5°C for the subsequent freshly poured RFC lift. (3) The hydration temperature rise was recorded the highest (13–14°C) in the areas near the dam abutment, due to challenges of placing rocks. These findings can provide insights and advice for similar construction scenarios.

Spatial–Temporal-Correlation-Constrained Dynamic Graph Convolutional Network for Traffic Flow Forecasting

Accurate traffic flow prediction in road networks is essential for intelligent transportation systems (ITS). Since traffic data are collected from the road network with spatial topological and time series sequences, the traffic flow prediction is regarded as a spatial–temporal prediction task. With the powerful ability to model the non-Euclidean data, the graph convolutional network (GCN)-based models have become the mainstream framework for traffic forecasting. However, existing GCN-based models either use the manually predefined graph structure to capture the spatial features, ignoring the heterogeneity of road networks, or simply perform 1-D convolution with fixed kernel to capture the temporal dependencies of traffic data, resulting in insufficient long-term temporal feature extraction. To solve those issues, a spatial–temporal correlation constrained dynamic graph convolutional network (STC-DGCN) is proposed for traffic flow forecasting. In STC-DGCN, a spatial–temporal embedding encoder module (STEM) is first constructed to encode the dynamic spatial relationships for road networks at different time steps. Then, a temporal feature encoder module with heterogeneous time series correlation modeling (TFE-HCM) and a spatial feature encoder module with dynamic multi-graph modeling (SFE-DCM) are designed to generate dynamic graph structures for effectively capturing the dynamic spatial and temporal correlations. Finally, a spatial–temporal feature fusion module based on a gating fusion mechanism (STM-GM) is proposed to effectively learn and leverage the inherent spatial–temporal relationships for traffic flow forecasting. Experimental results from three real-world traffic flow datasets demonstrate the superior performance of the proposed STC-DGCN compared with state-of-the-art traffic flow forecasting models.

Association of Prehospital Rearrest With Outcome Following Out-of-Hospital Cardiac Arrest: A Systematic Review and Meta-Analysis of Observational Studies

Objectives Exposure to prehospital rearrest has previously been associated with mortality following out-of-hospital cardiac arrest (OHCA). Our objective was to conduct a systematic review and meta-analysis examining the association between prehospital rearrest and survival in adults following OHCA resuscitation. Methods We searched the PubMed, Scopus, and Web of Science bibliographic databases for observational studies that included adult OHCA patients who achieved return of spontaneous circulation in the prehospital setting following OHCA and reported survival to hospital discharge data stratified by rearrest status. The primary exposure was prehospital rearrest. The primary outcome for this study was survival to hospital discharge. Secondary outcomes included survival with a favorable neurological outcome and rearrest prevalence. We pooled data using inverse heterogeneity modeling and presented effect sizes for the survival outcomes as odds ratios with 95% confidence intervals. We quantified heterogeneity using Cochran’s Q and the I2 statistic and examined small study effects using Doi plots and the LFK index. Results Of the 84 publications screened, we included 7 observational studies containing 27,045 patients with survival to hospital discharge data. Rearrest was common (30% [18–43%]; n = 7 studies; Q = 1086.1, p < 0.001; I2 = 99%; LFK index = 1.21) and associated with both decreased odds of survival to discharge (pooled aOR: 0.27 [0.22, 0.33]; n = 7 studies; Q = 32.2, p < 0.01, I2 = 81%, LFK index = −0.08) and decreased odds of survival to discharge with a favorable neurological outcome (pooled aOR: 0.25, [0.22, 0.28]; n = 4 studies; Q = 3.5, p = 0.3; I2 = 13%, LFK index = 1.30). Conclusions Rearrest is common and associated with decreased survival following OHCA. The pooled result of this meta-analysis suggests that preventing rearrest in five patients would be necessary to save one life.

Development of optimized ensemble machine learning-based prediction models for wire electrical discharge machining processes

This paper proposes development of optimized heterogeneous ensemble models for prediction of responses based on given sets of input parameters for wire electrical discharge machining (WEDM) processes, which have found immense applications in many of the present-day manufacturing industries because of their ability to generate complicated 2D and 3D profiles on hard-to-machine engineering materials. These ensembles are developed combining predictions of the three base models, i.e. random forest, support vector machine and ridge regression. These three base models are first framed utilizing the training datasets, providing predictions for all the responses under consideration. Based on these predictions, two optimization problems are formulated for each of the responses, while minimizing root mean squared error and mean absolute error, for subsequent development of two optimized ensembles whose predictions are the weighted sum of the predictions of the base models. The prediction performance of all the five models is ascertained through nine statistical metrics, after which a cumulative quality loss-based multi-response signal-to-noise (MRSN) ratio for each model is computed, for each of the responses, where a higher MRSN ratio indicates greater accuracy in prediction. This study is conducted using two experimental datasets of WEDM process. Overall, the optimized ensemble models having higher MRSN ratios than the base models are indicated to deliver better prediction accuracy.

Application of first- and second-order adjoint methods to glacial isostatic adjustment incorporating rotational feedbacks

SUMMARY This paper revisits and extends the adjoint theory for glacial isostatic adjustment (GIA) of Crawford et al. (2018). Rotational feedbacks are now incorporated, and the application of the second-order adjoint method is described for the first time. The first-order adjoint method provides an efficient means for computing sensitivity kernels for a chosen objective functional, while the second-order adjoint method provides second-derivative information in the form of Hessian kernels. These latter kernels are required by efficient Newton-type optimization schemes and within methods for quantifying uncertainty for non-linear inverse problems. Most importantly, the entire theory has been reformulated so as to simplify its implementation by others within the GIA community. In particular, the rate-formulation for the GIA forward problem introduced by Crawford et al. (2018) has been replaced with the conventional equations for modelling GIA in laterally heterogeneous earth models. The implementation of the first- and second-order adjoint problems should be relatively easy within both existing and new GIA codes, with only the inclusions of more general force terms being required.

Consumption tax cuts vs stimulus payments

Recent work shows that in macroeconomic models with non-Ricardian consumer behavior, uniform transfers are equivalent to interest rate cuts. That is, policymakers can use stimulus payments to substitute for conventional monetary policy when, say, the zero lower bound on short-term rates binds. We argue that in the same class of models, temporarily reducing consumption taxes delivers more stimulus than transfers — at the same cost to the taxpayer. Consumption tax cuts activate both income and substitution channels and prompt a strong response from all consumers across the wealth distribution. Simulating these policies in a quantitative heterogeneous agent New Keynesian model, we find output expands twice as much.

Cellular liquid biopsy provides unique chances for disease monitoring, preclinical model generation and therapy adjustment in rare salivary gland cancer patients.

While cell-free liquid biopsy (cfLB) approaches provide simple and inexpensive disease monitoring, cell-based liquid biopsy (cLB) may enable additional molecular genetic assessment of systemic disease heterogeneity and preclinical model development. We investigated 71 blood samples of 62 patients with various advanced cancer types and subjected enriched circulating tumor cells (CTCs) to organoid culture conditions. CTC-derived tumoroid models were characterized by DNA/RNA sequencing and immunohistochemistry, as well as functional drug testing. Results were linked to molecular features of primary tumors, metastases, and CTCs; CTC enumeration was linked to disease progression. Of 52 samples with positive CTC counts (≥1) from eight different cancer types, only CTCs from two salivary gland cancer (SGC) patients formed tumoroid cultures (P = 0.0005). Longitudinal CTC enumeration of one SGC patient closely reflected disease progression during treatment and revealed metastatic relapse earlier than clinical imaging. Multiomics analysis and functional invitro drug testing identified potential resistance mechanisms and drug vulnerabilities. We conclude that cLB might add a functional dimension (to the genetic approaches) in the personalized management of rare, difficult-to-treat cancers such as SGC.

Deciphering autism heterogeneity: a molecular stratification approach in four mouse models

Autism spectrum disorder (ASD) is a complex neurodevelopmental condition characterized by impairments in social interaction and communication, as well as restrained or stereotyped behaviors. The inherent heterogeneity within the autism spectrum poses challenges for developing effective pharmacological treatments targeting core features. Successful clinical trials require the identification of robust markers to enable patient stratification. In this study, we identified molecular markers within the oxytocin and immediate early gene families across five interconnected brain structures of the social circuit. We used wild-type and four heterogeneous mouse models, each exhibiting unique autism-like behaviors modeling the autism spectrum. While dysregulations in the oxytocin family were model-specific, immediate early genes displayed widespread alterations, reflecting global changes across the four models. Through integrative analysis, we identified Egr1, Foxp1, Homer1a, Oxt, and Oxtr as five robust and discriminant molecular markers that allowed the successful stratification of the four models. Importantly, our stratification demonstrated predictive values when challenged with a fifth mouse model or identifying subgroups of mice potentially responsive to oxytocin treatment. Beyond providing insights into oxytocin and immediate early gene mRNA dynamics, this proof-of-concept study represents a significant step toward the potential stratification of individuals with ASD. This work has implications for the success of clinical trials and the development of personalized medicine in autism.

Modelling of a continuous sorption-enhanced methanation process in an adiabatic packed-bed reactor system

The present study investigates a sorption-enhanced methanation process and successive solid regeneration stages using a dual-function catalyst containing nickel as a catalyst and zeolite 13X for water removal. An adiabatic packed bed, modelled by a dynamical and heterogeneous model, has been considered, and a five-stage sequence describes the sorption-enhanced methanation and successive solid regeneration. First, methanation occurs with in-situ water removal; then, the catalyst drying using a pressure and temperature swing approach implemented by blowdown, purge, cooling, and pressurisation stages. In adiabatic operation, heat management is crucial; on the other hand, the heat produced can be efficiently used to dry the zeolite. Process intensification is pursued by addressing the effect of gas inlet temperature, pressure, and GHSV on system performance. Achieving an average purity of 99 % of the methane for pressures greater than 2 bar is possible for long-time operation, with productivity averaging 0.8 mol/(kgads min) at the highest gas hourly space velocity investigated.

Origin of ventricular fibrillation triggers in a model of localized repolarization heterogeneity

BackgroundHeterogeneities in ventricular repolarization contribute significantly to the genesis of ventricular fibrillation (VF). Although clinical arrhythmias are spontaneously triggered by premature ventricular complexes, these triggers are difficult to document and little is known about their site of origin. ObjectiveTo characterize spontaneous VF initiation in an experimental model of repolarization heterogeneity and to identify the origin of triggers in relation to the spatial dispersion of repolarization. MethodsSpatially limited repolarization heterogeneity was created in isolated perfused porcine right ventricles (N =16) by local administration of pinacidil (20 μM) into a terminal branch of the right coronary artery. High-resolution optical mapping and pseudo-ECG were performed in control conditions and after pinacidil perfusion. ResultsNo arrhythmia occurred at baseline, but 74 VF episodes were observed in 13 hearts (82%) after pinacidil perfusion and were most often initiated by a ventricular trigger with short coupling interval (297±66ms). Sixteen VF initiations were optically mapped in four hearts. Mapping showed triggers originating in all cases from the border zone between altered and normal repolarization areas where local action potential duration and repolarization time gradients were steep (15.9 and 15.8 ms/mm versus 1.5 and 3.0 ms/mm in non-trigger sites). Optical action potential traces were compatible with a phase-two reexcitation mechanism. The subsequent VF cycles were driven by activities located in the same region. ConclusionsThis model of localized repolarization heterogeneity is able to produce spontaneous VF initiation. Our study demonstrates that VF triggers originate consistently from the border zone of repolarization dispersion.

A heterogeneous continuum traffic flow model considering mixed connected and autonomous vehicles

Under the background of mixed scenarios of Connected and Autonomous Vehicles and Connected and Autonomous Electric Vehicles, a novel mixed car-following model considering the information of multiple vehicles and the difference of safety spacing in different driving scenarios is proposed. Then, the corresponding heterogeneous continuum traffic flow model is derived. Secondly, through the linear and nonlinear analysis approaches, the linear stability condition and the KdV-Burgers equation are yielded to describe the characteristics of traffic flow evolution. Finally, numerical simulations are performed to explore the density evolution of traffic flow under different initial density and permeability conditions. The conclusions can be summarized as follows: (1) The derived model can well simulate spatiotemporal phase changes and traffic flow phenomena such as stop-and-go waves and local cluster effects. (2) With the increase of the permeability of electric vehicles and speed difference sensitivity coefficient, the road density decreases from 0.15veh/m to 0.06veh/m, the traffic congestion area decreases, and the traffic congestion is effectively suppressed, which also verifies the rationality of the above theory.

FedDA: Resource-adaptive federated learning with dual-alignment aggregation optimization for heterogeneous edge devices

Federated learning (FL) is an emerging distributed learning paradigm that allows multiple clients to collaborate on training a global model without sharing their local data. However, in practical heterogeneous edge device scenarios, FL faces the challenges of system heterogeneity and data heterogeneity, which leads to unfair participation and degraded global model performance. In this paper, we introduce FedDA, a resource-adaptive FL framework, which adapts to the client’s computing resources by assigning heterogeneous models of different sizes. To improve the performance of heterogeneous model aggregation and adjust to non-independent and identically distributed (non-i.i.d.) data, we propose a dual-alignment aggregation optimization method, i.e., parameter feature space alignment and output space alignment. Specifically, FedDA exploits the permutation symmetry property of weight space to permutate the model parameter positions via an adaptive layer-wise matching method, thereby aligning models with significant deviations in parameter feature space. FedDA mitigates the imbalance in parameter quantity between smaller and larger models through parameter expansion. Additionally, FedDA maps client labels into a uniform embedding space through output space alignment, thus reducing model parameter deviations due to non-i.i.d. data without additional client-side computing overhead. We evaluate the performance of FedDA on benchmark datasets, including FashionMNIST, CIFAR10, CIFAR100 and AGNews. Experimental results demonstrate that FedDA achieves up to 8.71% improvement in model accuracy compared to baseline methods, highlighting its effectiveness in addressing the challenges of heterogeneity.

Visualized analysis of microscale rock mechanism research: A bibliometric data mining approach

Rock mechanics is an indispensable discipline in diverse sectors, from resource retrieval to disaster mitigation. Diving deeper into this field, particularly into microscale rock mechanics, offers strategic insights and potential advancements for rock engineering practice. The objective of this research is to map the scientific production tied to microscale rock mechanics to date. In doing so, we perform a bibliometric analysis to look over and discuss the performance of the related literature. The conceptual fabric of microscale rock mechanics research, constituted by four central themes, was revealed and visualized through a text mining analysis. These areas include (1) modeling and simulation of fluid flow and transport within porous media, (2) characterizing fracture and failure in rocks, (3) understanding the deformation mechanism in response to geological processes, and (4) studying the mechanical properties of rocks subjected to extreme conditions. Lastly, an upcoming research agenda for microscale rock mechanics was proposed, centered on addressing three identified research gaps: (I) the integration of geological processes to characterize mineral properties, (II) the augmentation of fracture predictions achieved through multiscale modeling of rock heterogeneity, and (III) the exploitation of Artificial Intelligence technologies for anticipating complex fracturing scenarios. This comprehensive approach promises to enrich our understanding of rock mechanics and its applications.

Co-seismic landslide susceptibility mapping for the Luding earthquake area based on heterogeneous ensemble machine learning models

ABSTRACT Considering the complexity of landslide susceptibility mapping (LSM) at the municipal-scale, it is difficult for the assumption space of a single machine learning (ML) algorithm to fully and accurately reflect the intrinsic correlation between the landslide influencing factors and the actual catastrophic events under all scenarios. This study addresses this by integrating three heterogeneous ensemble learning (HEL) algorithms(Bagging, Blending, and Stacking) with four ML models (Gradient Boosting Decision Tree (GBDT), Extreme Gradient Boosting (XGBoost), K-Nearest Neighbors (KNN), and Backpropagation-Neural Network (BPNN)), to construct 27 HEL LSM models, and to realize optimization and selection of the models and algorithms through the comparative analysis of the results of the LSM and the accuracy of the prediction of the individual models. The final results indicate that the Bagging and Blending HEL algorithms fail to enhance model prediction accuracy effectively. In contrast, the Stacking HEL algorithm significantly improves model prediction accuracy (with an accuracy increase of over 1.4% and AUC values above 0.95). The Stacking (GBDT-XGBoost) model had the highest prediction accuracy (AUC = 0.964) and thus was selected as the optimal LSM model. This study's modeling and mapping results are intended to serve as a reference for geohazard prevention and sustainability studies in related cities.

Device-Performance-Driven Heterogeneous Multiparty Learning for Arbitrary Images.

Multiparty learning (MPL) is an emerging framework for privacy-preserving collaborative learning. It enables individual devices to build a knowledge-shared model and remaining sensitive data locally. However, with the continuous increase of users, the heterogeneity gap between data and equipment becomes wider, which leads to the problem of model heterogeneous. In this article, we concentrate on two practical issues: data heterogeneous problem and model heterogeneous problem, and propose a novel personal MPL method named device-performance-driven heterogeneous MPL (HMPL). First, facing the data heterogeneous problem, we focus on the problem of various devices holding arbitrary data sizes. We introduce a heterogeneous feature-map integration method to adaptively unify the various feature maps. Meanwhile, to handle the model heterogeneous problem, as it is essential to customize models for adapting to the various computing performances, we propose a layer-wise model generation and aggregation strategy. The method can generate customized models based on the device's performance. In the aggregation process, the shared model parameters are updated through the rules that the network layers with the same semantics are aggregated with each other. Extensive experiments are conducted on four popular datasets, and the result demonstrates that our proposed framework outperforms the state of the art (SOTA).

Influence of financial inclusion and industrial robots on the efficiency of China's OFDI: A heterogeneous spatial stochastic frontier analysis

Enhancing the efficacy of China's outward foreign direct investment (COFDI) constitutes a significant avenue for realizing the sustainable development of China's economy. This study analyzes panel data from 63 countries from 2010 to 2021 to create a heterogeneous spatial stochastic frontier model to examine how financial inclusion and industrial robotics in host nations affect COFDI efficiency. The findings are as follows. To begin with, COFDI manifests positive spatial spillover effects across countries. Moreover, from a spatial perspective, financial inclusion diminishes COFDI efficiency, whereas the interaction between financial inclusion and industrial robots increases its efficiency. Industrial robots, thus, counteract the diminishing impact of financial inclusion on COFDI efficiency. Additionally, concerning the heterogeneity of location choice and investment motivation in COFDI, the interaction between financial inclusion and industrial robots in developed economies augments COFDI efficiency as opposed to emerging market economies. Conversely, technology-seeking investment mitigates the adverse effects of financial inclusion on the efficacy of COFDI when juxtaposed with resource-seeking investment. This study provides theoretical insights and practical recommendations for improving the efficiency and guiding the destination selection of COFDI.

Momentum mechanisms under heterogeneous beliefs

We establish a continuous-time heterogeneous beliefs model to discuss the mechanisms of Momentum and Reversal. Price learning, information transmission and extrapolative expectation are incorporated into a unified framework for the Momentum and Reversal. The calibration results from SP500 show that the presence of Extrapolators and Information-driven Traders are important influences of Momentum and Reversal in all phases. We also find that momentum and reversal become significantly stronger as belief weights approach true belief weights.