Model Setup Research Articles

The tumor stroma influences immune cell distribution and recruitment in a PDAC-on-a-chip model.

The dense tumor stroma of pancreatic ductal adenocarcinoma (PDAC) and its secreted immune active molecules provide a barrier for chemotherapy treatment as well as for immune cell infiltration to the tumor core, providing a challenge for immunotherapeutic strategies. Consequently, the investigation of processes underlying the interaction between the tumor stroma, particularly activated pancreatic stellate cells (PSCs), and immune cells may offer new therapeutic approaches for PDAC treatment. In this study, we established a 3D PDAC model cultured under flow, consisting of an endothelial tube, PSCs and PDAC organoids. This was applied to study the role of the tumor microenvironment (TME) on immune cell recruitment and its effect on partly preventing their interaction with pancreatic cancer cells. We observed that stromal cells form a physical barrier, partly shielding the cancer cells from migrating immune cells, as well as a biochemical microenvironment, that seems to attract and influence immune cell distribution. In addition, stromal targeting by Halofuginone led to an increase in immune cell infiltration. We propose that the here developed model setups will support the understanding of the cellular interplay influencing the recruitment and distribution of immune cells, and contribute to the identification of key players in the PDAC immunosuppressive TME as well as support the discovery of new strategies to treat this immune unresponsive tumor.

Toward nebular spectral modeling of magnetar-powered supernovae

Context. Many energetic supernovae (SNe) are thought to be powered by the rotational energy of a highly magnetized, rapidly rotating neutron star. The emission from the associated luminous pulsar wind nebula (PWN) can photoionize the SN ejecta, leading to a nebular spectrum of the ejecta with signatures that might reveal the PWN. SN 2012au is hypothesized to be one such SN. Aims. We investigate the impact of different ejecta and PWN parameters on the SN nebular spectrum, and test whether any photoionization models are consistent with SN 2012au. We study how constraints from the nebular phase can be linked into modeling of the diffusion phase and the radio emission of the magnetar. Methods. We present a suite of late-time (1–6 yr) spectral simulations of SN ejecta powered by an inner PWN. Over a large grid of one-zone models, we study the behavior of the physical state and line emission of the SN as the PWN luminosity (LPWN), the injected spectral energy distribution (SED) temperature (TPWN), the ejecta mass (Mej), and the composition (pure O or realistic) vary. We discuss the resulting emission in the context of the observed behavior of SN 2012au, a strong candidate for a PWN-powered SN. We used optical light-curve models and broadband PWN models to predict possible radio emission from SN 2012au. Results. The SN nebular spectrum varies as TPWN varies because the ejecta become less ionized as TPWN increases. Ejecta models with low mass and high PWN power obtain runaway ionization for O I, and in extreme cases, also O II, causing a sharp decrease in their ion fraction over a small change in the parameter space. Certain models can reproduce the oxygen line luminosities of SN 2012au reasonably well at individual epochs, but we find no model that fits over the whole time evolution. This is likely due to uncertainties and simplifications in the model setup. Using our derived constraints from the nebular phase, we predict that the magnetar powering SN 2012au had an initial rotation period ~15 ms, and it is expected to be a strong radio source (F > 100 μJy) for decades.

A Dimensionless Parameter for Predicting Convective Self‐Aggregation Onset in a Stochastic Reaction‐Diffusion Model of Tropical Radiative‐Convective Equilibrium

AbstractWe introduce a minimal stochastic lattice model for the column relative humidity (R) in the tropics, which incorporates convective moistening, horizontal transport and subsidence drying. The probability of convection occurring in a location increases with R, based on Tropical Rainfall Measuring Mission observations, providing a positive feedback that could lead to aggregation. We show that the simple model reproduces many aspects of full‐physics cloud resolving model experiments. Depending on model parameter settings and domain size and resolution choices, it can produce both random and aggregated equilibrium states. Clustering occurs more readily with larger domains and coarser resolutions, in agreement with full‐physics models. Using dimensional arguments and fits from empirical data, we derive a dimensionless parameter which we call the aggregation number, Nag, that predicts whether a specific model and experiment setup will result in an aggregated or random state. The parameter includes the moistening feedback strength, the horizontal moisture transport efficiency, the subsidence timescale, the domain size and spatial resolution. Using large ensembles of experiments, we show that the transition between random and aggregated states occurs at a critical value of Nag. We argue that Nag could help to understand the differences in aggregation states between full‐physics, cloud resolving models, which show little consensus about the robustness of self‐organized patterns, whose emergence appears to be sensitive to the model setup, physics and parameterizations.

High Reliable Medium Voltage Drive With Reduced Component Count of Converters

The rapid development of technology, particularly in the fields of medium-voltage drive applications, has led to the widespread and use of AC to DC multi-pulse and DC to AC multilevel converters. In recent years, multi-winding transformers (MWTs)-based multi-pulse converters have received significant attention, owing to the growing need to utilize the energy as per power quality standards with high reliability in medium voltage drive systems. The main issues of multi-pulse converters are the large complexity of winding connections, which directly degrades the reliability of the system and increases the requirement for conductor material. To address these issues of grid-end converters, this paper suggests a less complex structure of a multi-winding transformer (MWT) for a thirty-six pulse AC-DC converter, which demands the least conductor material for windings. Moreover, the reduction in the number of components of the multi-level inverters has a noticeable role in saving size, cost, and reliability improvement. To address these features in the drive system, a five-level hybrid symmetric cascaded H-bridge multilevel inverter (HSCHB-MLI) is presented, which performs a high-quality power conversion at the motor terminals of the drive system. Hence, the main concerns of this drive system are the reduction in the component count, weight, and losses of the system with excellent harmonics performance. The operational aspects of the topology such as circuit structure, design, and indirect field-oriented control (IFOC), and modulation technique (parabolic carrier wave-based level shifted pulse width modulation) are elaborated in detail. A detailed comparative assessment is performed with other multi-winding transformers and multi-level inverter structures to demonstrate the merits of the presented multi-winding transformer and multilevel inverter structures, respectively. Finlay, through Simulink modeling and hardware testing setup, the viability of the proposed architecture is verified.

A BIM-based framework for automatic numerical modelling and geotechnical analysis of a large-scale deep excavation for transportation infrastructures

Abstract Employing emerging information and communication technologies such as building information modelling (BIM) to streamline engineering design and analysis has been mainstream over the past several years. Although BIM models are well established for aboveground infrastructures, underground BIM is still in its infancy. On one hand, valuable subsurface ground models are missing in current BIM applications as site-specific geotechnical data are often sparse and limited. On the other hand, BIM has been mainly used as a visualization tool for concise representation of design and construction data, and it has not been integrated and utilized to assess geotechnical risks associated with underground infrastructures. This paper proposes a BIM-based approach for automatic numerical modelling and geotechnical analysis. Structural information (e.g. geometries) of different project BIMs are grouped, exported, and saved, which provides a unified interface for automatic information extraction using computer codes (e.g. Python). Subsequently, subsurface ground models generated from sparse data are integrated with extracted basic geometric properties for automatic geotechnical model setup and finite element analysis. The performance of the proposed framework is illustrated using a real deep excavation project. It is revealed that BIM, as a data repository, enables timely and accurate information exchange between structural and geotechnical models in an automatic manner, which emphasizes the need of a BIM-based approach for assessing and managing geotechnical risks associated with deep excavations for underground transportation infrastructures, especially in the urban context.

The impact of religious involvement on trust, volunteering, and perceived cooperativeness: evidence from two British panels

Abstract Does religious involvement make people more trusting and prosocial? Considering conflicting theories and mixed prior evidence, we subject this question to a stringent test using large-scale, representative data from the British Household Panel Survey (1991–2009, N ≈ 26,000) and the UK Household Longitudinal Study (2009–2021, N ≈ 80,000). We employ cross-lagged panel models with individual fixed effects to account for time-invariant confounders and reverse causality—two issues that have haunted earlier research. We find that frequency of religious service attendance on average has a positive impact on generalized trust, volunteering, and perceived cooperativeness. Other indicators of religious involvement have weaker effects. We also find variation across religious traditions: the effects of religious attendance are mostly positive for Anglicans and other Protestants, but weaker and mostly statistically insignificant for Catholics, Hindus, and the unaffiliated, and even negative for Muslims when the outcome is perceived cooperativeness. Our findings are robust to alternative model set-ups and hold up after accounting for neighbourhood religious composition, respondent and interviewer ethnicity, and other potential moderators and confounders. Altogether, our study shows that religious involvement can foster prosocial behaviours and attitudes, although in our study this effect is largely restricted to religious service attendance and majority religions.

Modeling Soil Organic Carbon Dynamics of Arable Land across Scales: A Simplified Assessment of Alternative Management Practices on the Level of Administrative Units

Regional assessments of soil organic carbon (SOC) trends and the carbon sequestration potential of alternative management practices (AMP) are highly relevant for developing climate change mitigation strategies for the agricultural sector. Such studies could benefit from simplified SOC modeling approaches on the scale of administrative units as this often corresponds to the level of policy-making and data availability. However, there is a risk of systematic errors in such scaling operations. To overcome this problem, we performed a scaling experiment where we simulated the SOC dynamics of the arable soils of the State of Saxony (Germany) across a series of scales using the CANDY Carbon Balance (CCB) model. Specifically, we developed model set-ups on four different administrative levels (NUTS1, NUTS2, NUTS3, and LAU) and evaluated the simulation results of the upscaled models against a 500 m grid-based reference model. Furthermore, we quantified the carbon sequestration potential of selected AMP scenarios (addressing field grass, cover crops, and conservation tillage) across all scales. The upscaled model set-ups adequately simulated the SOC trends of Saxon arable land compared to the grid-based reference simulation (scaling error: 0.8–3.8%), while providing significant benefits for model application, data availability and runtime. The carbon sequestration potential of the AMP scenarios (1.33 Mt C until 2050) was slightly overestimated (+0.07–0.09 Mt C) by the upscaled model set-ups. Regardless of the scale of model set-up, we showed that the use of aggregated statistical input data could lead to a systematic underestimation of SOC trends. LAU and NUTS3 levels were shown to be a suitable compromise for effectively quantifying SOC dynamics and allowed for an acceptable spatial prioritization of AMPs. Such simplified, scale-adapted assessments are valuable for cross-regional comparisons and for communication to and among decision-makers, and might provide a quantitative basis for discussions on the effectiveness of AMPs in various stakeholder processes.

Dynamic finite element analysis of hip replacement edge loading: Balancing precision and run time in a challenging model

An important aspect in evaluating the resilience of hip replacement designs is testing their performance under adverse conditions that cause edge loading of the acetabular liner. The representation of edge loading conditions in finite element models is computationally challenging due to the changing contact locations, need for fine meshes, and dynamic nature of the system.In this study, a combined mesh and mass-scaling sensitivity study was performed to identify an appropriate compromise between convergence and solution time of explicit finite element analysis in investigating edge loading in hip replacement devices. The optimised model was then used to conduct a sensitivity test investigating the effect of different hip simulator features (the mass of the translating fixture and mediolateral spring damping) on the plastic strain in the acetabular liner. Finally, the effect of multiple loading cycles on the progressive accumulation of plastic strain was then also examined using the optimised model.A modelling approach was developed which provides an effective compromise between mass-scaling effects and mesh refinement for a solution time per cycle of less than 1 h. This ‘Recommended Mesh’ model underestimated the plastic strains by less than 10%, compared to a ‘Best Estimate’ model with a run time of ∼190 h. Starting with this model setup would therefore significantly reduce any new model development time while also allowing the flexibility to incorporate additional complexities as required. The polyethylene liner plastic strain was found to be sensitive to the simulator mass and damping (doubling the mass or damping had a similar magnitude effect to doubling the swing phase load) and these should ideally be described in future experimental studies. The majority of the plastic strain (99%) accumulated within the first three load cycles.

Foundations and guidelines for high-quality three-dimensional models using photogrammetry: A technical note on the future of neuroanatomy education.

Hands-on dissections using cadaveric tissues for neuroanatomical education are not easily available in many educational institutions due to financial, safety, and ethical factors. Supplementary pedagogical tools, for instance, 3D models of anatomical specimens acquired with photogrammetry are an efficient alternative to democratize the 3D anatomical data. The aim of this study was to describe a technical guideline for acquiring realistic 3D anatomic models with photogrammetry and to improve the teaching and learning process in neuroanatomy. Seven specimens with different sizes, cadaveric tissues, and textures were used to demonstrate the step-by-step instructions for specimen preparation, photogrammetry setup, post-processing, and display of the 3D model. The photogrammetry scanning consists of three cameras arranged vertically facing the specimen to be scanned. In order to optimize the scanning process and the acquisition of optimal images, high-quality 3D models require complex and challenging adjustments in the positioning of the specimens within the scanner, as well as adjustments of the turntable, custom specimen holders, cameras, lighting, computer hardware, and its software. MeshLab® software was used for editing the 3D model before exporting it to MedReality® (Thyng, Chicago, IL) and SketchFab® (Epic, Cary, NC) platforms. Both allow manipulation of the models using various angles and magnifications and are easily accessed using mobile, immersive, and personal computer devices free of charge for viewers. Photogrammetry scans offer a 360° view of the 3D models ubiquitously accessible on any device independent of operating system and should be considered as a tool to optimize and democratize the teaching of neuroanatomy.

Sensitivity of WRF/Chem simulated PM2.5 to initial/boundary conditions and planetary boundary layer parameterization schemes over the Indo-Gangetic Plain.

The ability of a chemical transport model to simulate accurate meteorological and chemical processes depends upon the physical parametrizations and quality of meteorological input data such as initial/boundary conditions. In this study, weather research and forecasting model coupled with chemistry (WRF-Chem) is used to test the sensitivity of PM2.5 predictions to planetary boundary layer (PBL) parameterization schemes (YSU, MYJ, MYNN, ACM2, and Boulac) and meteorological initial/boundary conditions (FNL, ERA-Interim, GDAS, and NCMRWF) over Indo-Gangetic Plain (Delhi, Punjab, Haryana, Uttar Pradesh, and Rajasthan) during the winter period (December 2017 to January 2018). The aim is to select the model configuration for simulating PM2.5 which shows the lowest errors and best agreement with the observed data. The best results were achieved with initial/boundary conditions from ERA and GDAS datasets and local PBL parameterization (MYJ and MYNN). It was also found that PM2.5 concentrations are relatively less sensitive to changes in initial/boundary conditions but in contrast show a stronger sensitivity to changes in the PBL scheme. Moreover, the sensitivity of the simulated PM2.5 to the choice of PBL scheme is more during the polluted hours of the day (evening to early morning), while that to the choice of the meteorological input data is more uniform and subdued over the day. This work indicates the optimal model setup in terms of choice of initial/boundary conditions datasets and PBL parameterization schemes for future air quality simulations. It also highlights the importance of the choice of PBL scheme over the choice of meteorological data set to the simulated PM2.5 by a chemical transport model.

A Stochastic Covariance Shrinkage Approach in Ensemble Transform Kalman Filtering

The Ensemble Kalman Filters (EnKF) employ a Monte-Carlo approach to represent covariance information, and are affected by sampling errors in operational settings where the number of model realizations is much smaller than the model state dimension. To alleviate the effects of these errors EnKF relies on model-specific heuristics such as covariance localization, which takes advantage of the spatial locality of correlations among the model variables. This work proposes an approach to alleviate sampling errors that utilizes a locally averaged-in-time dynamics of the model, described in terms of a climatological covariance of the dynamical system. We use this covariance as the target matrix in covariance shrinkage methods, and develop a stochastic covariance shrinkage approach where synthetic ensemble members are drawn to enrich both the ensemble subspace and the ensemble transformation. We additionally provide for a way in which this methodology can be localized similar to the state-of-the-art LETKF method, and that for a certain model setup, our methodology significantly outperforms it.

A Transient 3D CFD Thermal Model of the Complete DI Diesel Engine Fuel System

<div class="section abstract"><div class="htmlview paragraph">This paper reports on a transient, three-dimensional computational fluid dynamics (CFD) study of flow and heat transfer in the complete fuel system of an inline 6-cylinder, direct injection (DI) diesel engine used in commercial applications. The CFD software Simerics-MP+ was used for this purpose. Diesel engine development, to meet fuel economy and exhaust emission standards, requires the precise integration of each component in the fuel system in order to reliably deliver the fuel to the combustion chamber as a function of crank angle to the combustion chamber, at the specified injection pressure. Both the model set-up and run times are practical, thus the simulation tool can play a key role in the design and development of diesel engine fuel systems.</div></div>

Multiscale analysis of nonlinear systems using a hierarchy of deep neural networks

The application of multiscale methods that are based on computational homogenization, such as the well established FE2, remains in most cases a computationally challenging task. In a system that involves more than two length scales, the task of a FEN (N¿2) calculation becomes practically intractable if applied in the frame of conventional computational setups. This paper addresses this issue by introducing a novel surrogate model designed to accelerate the solution procedure of hierarchically formulated multiscale problems. The idea is to employ a sequence of deep feedforward neural networks (DNNs) that represent the hierarchy of the separate scales in the multiscale problem. Each DNN is trained to learn the constitutive law of a corresponding length scale of the problem. In a similar manner to the original problem, where each finer scale is contained in a coarser scale, DNNs representing fine scales are contained in the DNNs that represent coarser scales. At the end of the training process, a single deep network is produced which emulates the macroscopic behaviour by incorporating all physical mechanisms arising at each of the problem’s finer scales. This approach takes full advantage of the accuracy and modelling capabilities of the FEN schemes, while at the same time handles the immense computational requirements associated with their implementation. By virtue of the elaborated surrogate modelling setup, a four-scale structural system is solved using a FE4 scheme at reasonable computational times. In turn, this allows us to perform laborious sensitivity analyses in order to assess the uncertainty in the material parameters and its propagation to the macroscopic structural response.

Mean Arterial Pressure and Cerebral Hemodynamics Across The Lifespan: A Cross-Sectional Study From Human Connectome Project-Aging.

Cerebral perfusion is directly affected by systemic blood pressure, which has been shown to be negatively correlated with cerebral blood flow (CBF). The impact of aging on these effects is not fully understood. To determine whether the relationship between mean arterial pressure (MAP) and cerebral hemodynamics persists throughout the lifespan. Retrospective, cross-sectional study. Six hundred and sixty-nine participants from the Human Connectome Project-Aging ranging between 36 and 100+ years and without a major neurological disorder. Imaging data was acquired at 3.0 Tesla using a 32-channel head coil. CBF and arterial transit time (ATT) were measured by multi-delay pseudo-continuous arterial spin labeling. The relationships between cerebral hemodynamic parameters and MAP were evaluated globally in gray and white matter and regionally using surface-based analysis in the whole group, separately within different age groups (young: <60 years; younger-old: 60-79 years; oldest-old: ≥80 years). Chi-squared, Kruskal-Wallis, ANOVA, Spearman rank correlation and linear regression models. The general linear model setup in FreeSurfer was used for surface-based analyses. P < 0.05 was considered significant. Globally, there was a significant negative correlation between MAP and CBF in both gray (ρ = -0.275) and white matter (ρ = -0.117). This association was most prominent in the younger-old [gray matter CBF (β = -0.271); white matter CBF (β = -0.241)]. In surface-based analyses, CBF exhibited a widespread significant negative association with MAP throughout the brain, whereas a limited number of regions showed significant prolongation in ATT with higher MAP. The associations between regional CBF and MAP in the younger-old showed a different topographic pattern in comparison to young subjects. These observations further emphasize the importance of cardiovascular health in mid-to-late adulthood for healthy brain aging. The differences in the topographic pattern with aging indicate a spatially heterogeneous relationship between high blood pressure and CBF. 3 TECHNICAL EFFICACY STAGE: 3.

Stochastic comparisons of largest claim and aggregate claim amounts

AbstractIn this paper, we establish some stochastic comparison results for largest claim amounts of two sets of independent and also for interdependent portfolios under the setup of the proportional odds model. We also establish stochastic comparison results for aggregate claim amounts of two sets of independent portfolios. Further, stochastic comparisons for largest claim amounts from two sets of independent multiple-outlier claims have also been studied. The results we obtained apply to the whole family of extended distributions, also known as the Marshall–Olkin family of distributions. We have given many numerical examples to illustrate the results obtained.

Reducing Southern Ocean Shortwave Radiation Errors in the ERA5 Reanalysis with Machine Learning and 25 Years of Surface Observations

Abstract Earth system models struggle to simulate clouds and their radiative effects over the Southern Ocean, partly due to a lack of measurements and targeted cloud microphysics knowledge. We have evaluated biases of downwelling shortwave radiation in the ERA5 climate reanalysis using 25 years (1995–2019) of summertime surface measurements, collected on the Research and Supply Vessel (RSV) Aurora Australis, the Research Vessel (R/V) Investigator, and at Macquarie Island. During October–March daylight hours, the ERA5 simulation of SWdown exhibited large errors (mean bias = 54 W m−2, mean absolute error = 82 W m−2, root-mean-square error = 132 W m−2, and R2 = 0.71). To determine whether we could improve these statistics, we bypassed ERA5’s radiative transfer model for SWdown with machine learning–based models using a number of ERA5’s gridscale meteorological variables as predictors. These models were trained and tested with the surface measurements of SWdown using a 10-fold shuffle split. An extreme gradient boosting (XGBoost) and a random forest–based model setup had the best performance relative to ERA5, both with a near complete reduction of the mean bias error, a decrease in the mean absolute error and root-mean-square error by 25% ± 3%, and an increase in the R2 value of 5% ± 1% over the 10 splits. Large improvements occurred at higher latitudes and cyclone cold sectors, where ERA5 performed most poorly. We further interpret our methods using Shapley additive explanations. Our results indicate that data-driven techniques could have an important role in simulating surface radiation fluxes and in improving reanalysis products. Significance Statement Simulating the amount of sunlight reaching Earth’s surface is difficult because it relies on a good understanding of how much clouds absorb and scatter sunlight. Relative to summertime surface observations, the ERA5 reanalysis still overestimates the amount of sunlight entering the Southern Ocean. We taught some models how to predict the amount of sunlight entering the Southern Ocean using 25 years of surface observations and a small set of meteorological variables from ERA5. By bypassing the ERA5’s internal simulation of the absorption and scattering of sunlight, we can drastically reduce biases in the predicted surface shortwave radiation. Large improvements in cold sectors of cyclones and closer to Antarctica were observed in regions where many numerical models struggle to simulate the amount of incoming sunlight correctly.

Sputum Culture Conversion among Patients with Drug-Resistant Tuberculosis: A Study of Various Predictors

Introduction: Conversion of a positive sputum culture to a negative culture is an important parameter in the management of drug-resistant tuberculosis (TB), as the failure to convert was related to dismal posttreatment outcomes and elevated mortality rates by almost eight times when compared to victims of multidrug-resistant TB (MDR-TB) who did so. In the patients of MDR-TB, it is an important sign of treatment success to have sputum culture converted from positive to negative, and helps in determining the duration of treatment. The goal of this study was to examine and compare the length of time it took to convert an initial sputum culture with predictors of treatment outcomes for patients with MDR-TB, as well as to pinpoint the variables that affected the length of time. Materials and Methods: The mixed method (records review and qualitative approach) was used to understand the patient’s perspective for accepting the Programmatic Management of Drug-resistant TB (PMDT) services and the provider’s perspective for rendering services in a public-private mix model setup. This study was undertaken as a retrospective analysis and studied the patients drug-resistant TB who were admitted to a tertiary care hospital in central India. This study was carried out on the patients of Rifampicin-Resistant TB (RR-TB) and MDR-TB (Resistant to both INH and Rifampicin) with positive baseline culture, pretreatment evaluation, and appropriate anti-MDR-TB treatment. As per the PMDT guidelines, sputum smear and culture reports were collected and analyzed. Results: In this study, the time and predictors to convert an initial positive sputum culture to negative and factors influencing treatment outcome were evaluated, compared, and identified. A total of 163 patients were evaluated and analyzed. Of these, 111 (68.09%) were rifampicin and INH-resistant (MDR-TB), whereas 52 (31.90%) were rifampicin resistant (RR-TB) only. Overall 68 (41.71%) patients with baseline positive culture were converted within 3 months. Conclusions: It was concluded that malnourishment, human immunodeficiency virus positivity, associated extra pulmonary TB, and bilateral cavitation were found to be significant predictors of longer time of sputum culture conversion and thereby treatment outcome.

Accelerated offline setup of homogenized microscopic model for multi‐scale analyses using neural network with knowledge transfer

AbstractThe FE computational homogenization method is a predictive multi‐scale method without the need for constitutive assumptions and/or potential function postulates at the macro engineering scale. Instead, the effective micro‐structural responses are extracted directly from a representative volume element (RVE) underlying each macro point. However, the FE method is still computationally too expensive for most practical uses, since the micro‐macro FE coupling is done at each loading step/iteration for the entire domain. To this end, the machine learning method has been utilized in the literature for the offline training of a surrogate model to predict the RVE homogenized response for general loading conditions. In this contribution, the neural network (NN) is incorporated into the macro finite element framework in a non‐intrusive manner. This is termed as the FE‐NN framework, in analogy to the FE method. In general, online simulations in the FE‐NN method is very efficient, with predictions matching closely to those obtained from reference direct numerical simulations (DNS). A bottleneck with the FE‐NN framework, however, is the high computational cost associated with the data generation for offline NN model setup. In this paper, focusing on the FE‐NN multi‐scale framework for non‐linear elastic deformation of heterogeneous materials, a sequential training strategy with knowledge transfer is proposed, to enable an efficient offline microscopic NN model setup. For a given target RVE, we first consider a simplified source RVE, where data can be generated rapidly, for the NN pre‐training of surrogate model. The pre‐trained network parameters are next downloaded to initialize the target NN surrogate model, followed by a fine‐tuning training process, using only a small dataset generated by the computationally expensive high‐fidelity RVE. The efficiency of the proposed sequential learning method over the conventional NN training, as well as, its excellent predictive capability for multi‐scale analyses, are demonstrated for a multi‐phase composite material. The proposed FE‐NN‐KT approach can be implemented easily without complicated pre‐processing procedures, since the data generation only involves the standard extraction of effective responses from RVEs, together with a standard NN architecture.

A genetic algorithm-based design model to provide reduced risk areas for housing interiors

PurposeThis study aims to present a novel Genetic Algorithm-Based Design Model (GABDM) to provide reduced-risk areas, namely, a “safe footprint,” in interior spaces during earthquakes. This study focuses on housing interiors as the space where inhabitants spend most of their daily lives.Design/methodology/approachThe GABDM uses the genetic algorithm as a method, the Nondominated Sorting Genetic Algorithm II algorithm, and the Wallacei X evolutionary optimization engine. The model setup, including inputs, constraints, operations and fitness functions, is presented, as is the algorithmic model’s running procedure. Following the development phase, GABDM is tested with a sample housing interior designed by the authors based on the literature related to earthquake risk in interiors. The implementation section is organized to include two case studies.FindingsThe implementation of GABDM resulted in optimal “safe footprint” solutions for both case studies. However, the results show that the fitness functions achieved in Case Study 1 differed from those achieved in Case Study 2. Furthermore, Case Study 2 has generated more successful (higher ranking) “safe footprint” alternatives with its proposed furniture system.Originality/valueThis study presents an original approach to dealing with earthquake risks in the context of interior design, as well as the development of a design model (GABDM) that uses a generative design method to reduce earthquake risks in interior spaces. By introducing the concept of a “safe footprint,” GABDM contributes explicitly to the prevention of earthquake risk. GABDM is adaptable to other architectural typologies that involve footprint and furniture relationships.

A comparison of various feature extraction and machine learning methods for antimicrobial resistance prediction in streptococcus pneumoniae

Streptococcus pneumoniae is one of the major concerns of clinicians and one of the global public health problems. This pathogen is associated with high morbidity and mortality rates and antimicrobial resistance (AMR). In the last few years, reduced genome sequencing costs have made it possible to explore more of the drug resistance of S. pneumoniae, and machine learning (ML) has become a popular tool for understanding, diagnosing, treating, and predicting these phenotypes. Nucleotide k-mers, amino acid k-mers, single nucleotide polymorphisms (SNPs), and combinations of these features have rich genetic information in whole-genome sequencing. This study compares different ML models for predicting AMR phenotype for S. pneumoniae. We compared nucleotide k-mers, amino acid k-mers, SNPs, and their combinations to predict AMR in S. pneumoniae for three antibiotics: Penicillin, Erythromycin, and Tetracycline. 980 pneumococcal strains were downloaded from the European Nucleotide Archive (ENA). Furthermore, we used and compared several machine learning methods to train the models, including random forests, support vector machines, stochastic gradient boosting, and extreme gradient boosting. In this study, we found that key features of the AMR prediction model setup and the choice of machine learning method affected the results. The approach can be applied here to further studies to improve AMR prediction accuracy and efficiency.