Predictive Capability Of Model Research Articles

Machine-Learning Prediction of Curie Temperature from Chemical Compositions of Ferromagnetic Materials.

Room-temperature ferromagnets are high-value targets for discovery given the ease by which they could be embedded within magnetic devices. However, the multitude of potential interactions among magnetic ions and their surrounding environments renders the prediction of thermally stable magnetic properties challenging. Therefore, it is vital to explore methods that can effectively screen potential candidates to expedite the discovery of novel ferromagnetic materials within highly intricate feature spaces. To this end, we explore machine-learning (ML) methods as a means to predict the Curie temperature (Tc) of ferromagnetic materials by discerning patterns within materials databases. This study emphasizes the importance of feature analysis and selection in ML modeling and demonstrates the efficacy of our gradient-boosted statistical feature-selection workflow for training predictive models. The models are fine-tuned through Bayesian optimization, using features derived solely from the chemical compositions of the materials data, before the model predictions are evaluated against literature values. We have collated ca. 35,000 Tc values and the performance of our workflow is benchmarked against state-of-the-art algorithms, the results of which demonstrate that our methodology is superior to the majority of alternative methods. In a 10-fold cross-validation, our regression model realized an R2 of (0.92 ± 0.01), an MAE of (40.8 ± 1.9) K, and an RMSE of (80.0 ± 5.0) K. We demonstrate the utility of our ML model through case studies that forecast Tc values for rare-earth intermetallic compounds and generate magnetic phase diagrams for various chemical systems. These case studies highlight the importance of a systematic approach to feature analysis and selection in enhancing both the predictive capability and interpretability of ML models, while being devoid of potential human bias. They demonstrate the advantages of such an approach over a mere reliance on algorithmic complexity and a black-box treatment in ML-based modeling within the domain of computational materials science.

Predicting significant wave height in the South China Sea using the SAC-ConvLSTM model

IntroductionThe precise forecasting of Significant wave height(SWH) is vital to ensure the safety and efficiency of aquatic activities such as ocean engineering, shipping, and fishing.MethodsThis paper proposes a deep learning model named SAC-ConvLSTM to perform 24-hour prediction with the SWH in the South China Sea. The long-term prediction capability of the model is enhanced by using the attention mechanism and context vectors. The prediction ability of the model is evaluated by mean absolute error (MAE), root mean square error (RMSE), mean square error (MSE), and Pearson correlation coefficient (PCC).ResultsThe experimental results show that the optimal input sequence length for the model is 12. Starting from 12 hours, the SAC-ConvLSTM model consistently outperforms other models in predictive performance. For the 24-hour prediction, this model achieves RMSE, MAE, and PCC values of 0.2117 m, 0.1083 m, and 0.9630, respectively. In addition, the introduction of wind can improve the accuracy of wave prediction. The SAC-ConvLSTM model also has good prediction performance compared to the ConvLSTM model during extreme weather, especially in coastal areas.DiscussionThis paper presents a 24-hour prediction of SWH in the South China Sea. Through comparative validation, the SAC-ConvLSTM model outperforms other models. The inclusion of wind data enhances the model's predictive capability. This model also performs well under extreme weather conditions. In physical oceanography, variables related to SWH include not only wind but also other factors such as mean wave period and sea surface air pressure. In the future, additional variables can be incorporated to further improve the model's predictive performance.

Integrated Single-cell and Bulk RNA Sequencing Analysis Cross Talk between Ferroptosis-related Genes and Prognosis in Oral Cavity Squamous Cell Carcinoma.

Ferroptosis is a new type of programmed apoptosis and plays an important role in tumour inhibition and immunotherapy. In this study, we aimed to explore the potential role of ferroptosis-related genes (FRGs) and the potential therapeutic targets in oral cavity squamous cell carcinoma (OCSCC). The transcription data of OCSCC samples were obtained from the Cancer Genome Atlas (TCGA) database as a training dataset. The prognostic FRGs were extracted by univariate Cox regression analysis. Then, we constructed a prognostic model using the least absolute shrinkage and selection operator (LASSO) and Cox analysis to determine the independent prognosis FRGs. Based on this model, risk scores were calculated for the OCSCC samples. The model's capability was further evaluated by the receiver operating characteristic curve (ROC). Then, we used the GSE41613 dataset as an external validation cohort to confirm the model's predictive capability. Next, the immune infiltration and somatic mutation analysis were applied. Lastly, single-cell transcriptomic analysis was used to identify the key cells. A total of 12 prognostic FRGs were identified. Eventually, 6 FRGs were screened as independent predictors and a prognostic model was constructed in the training dataset, which significantly stratified OCSCC samples into high-risk and low-risk groups based on overall survival. The external validation of the model using the GSE41613 dataset demonstrated a satisfactory predictive capability for the prognosis of OCSCC. Further analysis revealed that patients in the highrisk group had distinct immune infiltration and somatic mutation patterns from low-risk patients. Mast cell infiltrations were identified as prognostic immune cells and played a role in OCSCC partly through ferroptosis. We successfully constructed a novel 6 FRGs model and identified a prognostic immune cell, which can serve to predict clinical prognoses for OCSCC. Ferroptosis may be a new direction for immunotherapy of OCSCC.

Air quality forecasting using a spatiotemporal hybrid deep learning model based on VMD–GAT–BiLSTM

The precise forecasting of air quality is of great significance as an integral component of early warning systems. This remains a formidable challenge owing to the limited information of emission source and the considerable uncertainties inherent in dynamic processes. To improve the accuracy of air quality forecasting, this work proposes a new spatiotemporal hybrid deep learning model based on variational mode decomposition (VMD), graph attention networks (GAT) and bi-directional long short-term memory (BiLSTM), referred to as VMD–GAT–BiLSTM, for air quality forecasting. The proposed model initially employ a VMD to decompose original PM2.5 data into a series of relatively stable sub-sequences, thus reducing the influence of unknown factors on model prediction capabilities. For each sub-sequence, a GAT is then designed to explore deep spatial relationships among different monitoring stations. Next, a BiLSTM is utilized to learn the temporal features of each decomposed sub-sequence. Finally, the forecasting results of each decomposed sub-sequence are aggregated and summed as the final air quality prediction results. Experiment results on the collected Beijing air quality dataset show that the proposed model presents superior performance to other used methods on both short-term and long-term air quality forecasting tasks.

Forward and Inverse Modeling of Ice Sheet Flow Using Physics‐Informed Neural Networks: Application to Helheim Glacier, Greenland

AbstractPredicting the future contribution of the ice sheets to sea level rise over the next decades presents several challenges due to a poor understanding of critical boundary conditions, such as basal sliding. Traditional numerical models often rely on data assimilation methods to infer spatially variable friction coefficients by solving an inverse problem, given an empirical friction law. However, these approaches are not versatile, as they sometimes demand extensive code development efforts when integrating new physics into the model. Furthermore, this approach makes it difficult to handle sparse data effectively. To tackle these challenges, we use the Physics‐Informed Neural Networks (PINNs) to seamlessly integrate observational data and governing equations of ice flow into a unified loss function, facilitating the solution of both forward and inverse problems within the same framework. We illustrate the versatility of this approach by applying the framework to two‐dimensional problems on the Helheim Glacier in southeast Greenland. By systematically concealing one variable (e.g., ice speed, ice thickness, etc.), we demonstrate the ability of PINNs to accurately reconstruct hidden information. Furthermore, we extend this application to address a challenging mixed inversion problem. We show how PINNs are capable of inferring the basal friction coefficient while simultaneously filling gaps in the sparsely observed ice thickness. This unified framework offers a promising avenue to enhance the predictive capabilities of ice sheet models, reducing uncertainties, and advancing our understanding of poorly constrained physical processes.

Direct Measurements of Dust Settling Velocity Under Low‐Density Atmospheres Using Time‐Resolved Particle Image Velocimetry

AbstractDust dynamics influence planetary atmospheres. However, the settling velocity of dust—and thus its residence time in the atmosphere—is often mispredicted. Challenging, indirect experiments involving few ideal particles revealed that dust settling velocity deviates from Stokes' law under rarefied atmospheres. While useful, such experiments are inadequate to simulate more complex scenarios, including variable particles sizes and shapes. Here, we present direct measurements of settling velocity for spherical particles under Earth‐to‐Mars atmospheric pressures using time‐resolved particle image velocimetry (TR‐PIV), and validate their robustness with existing models. Our results demonstrate that TR‐PIV provides a relatively simple approach to quantifying dust settling velocity from direct observations of over 10,000 particles, enabling systematic investigations of dust settling under realistic scenarios. Such experiments will have significant implications for our understanding of Mars' past, present, and future ‐ from providing a tool to decipher its sedimentary record to enhancing predictive capabilities of atmospheric models.

Mathematical Modeling of Vedolizumab Treatment's Effect on Microbiota and Intestinal Permeability in Inflammatory Bowel Disease Patients.

Growing evidence suggests that impaired gut permeability and gut microbiota alterations are involved in the pathogenesis of Inflammatory Bowel Diseases (IBDs), which include Ulcerative Colitis (UC) and Crohn's Disease (CD). Vedolizumab is an anti-α4β7 antibody approved for IBD treatment, used as the first treatment or second-line therapy when the first line results in inadequate effectiveness. The aim of this study is to develop a mathematical model capable of describing the pathophysiological mechanisms of Vedolizumab treatment in IBD patients. In particular, the relationship between drug concentration in the blood, colonic mucosal permeability and fecal microbiota composition was investigated and modeled to detect and predict trends in order to support and tailor Vedolizumab therapies. To pursue this aim, clinical data from a pilot study on a cluster of 11 IBD patients were analyzed. Enrolled patients underwent colonoscopy in three phases (before (t0), after 24 weeks of (t1) and after 52 weeks of (t2 ) Vedolizumab treatment) to collect mucosal biopsies for transepithelial electrical resistance (TEER) evaluation (permeability to ions), intestinal permeability measurement and histological analysis. Moreover, fecal samples were collected for the intestinal microbiota analysis at the three time points. The collected data were compared to those of 11 healthy subjects at t0, who underwent colonoscopy for screening surveillance, and used to implement a three-compartmental mathematical model (comprising central blood, peripheral blood and the intestine). The latter extends previous evidence from the literature, based on the regression of experimental data, to link drug concentration in the peripheral blood compartment with Roseburia abundance and intestinal permeability. The clinical data showed that Vedolizumab treatment leads to an increase in TEER and a reduction in intestinal permeability to a paracellular probe, improving tissue inflammation status. Microbiota analysis showed increasing values of Roseburia, albeit not statistically significant. This trend was adequately reproduced by the mathematical model, which offers a useful tool to describe the pathophysiological effects of Vedolizumab therapy on colonic mucosal permeability and fecal microbiota composition. The model's satisfactory predictive capabilities and simplicity shed light on the relationship between the drug, the microbiota and permeability and allow for its straightforward extension to diverse therapeutic conditions.

Machine Learning Prediction of Small Molecule Accumulation in Escherichia Coli Enhanced with Descriptor Statistics.

Antibiotic resistance, particularly among Gram-negative bacteria, poses a significant healthcare challenge due to their ability to evade antibiotic action through various mechanisms. In this study, we explore the prediction of small molecule accumulation in Gram-negative bacteria by using machine learning techniques enhanced with statistical descriptors derived from molecular dynamics simulations. We begin by identifying a minimal set of molecular descriptors that maximize the model's predictive power while preserving human interpretability. We optimize model accuracy, precision, and the area under the receiver operating characteristic curve through an iterative process. We demonstrate that the inclusion of statistical descriptors significantly improves model performance across various prediction metrics. Particularly, the addition of statistical descriptors related to dipole moment and minimum projection radius enhances the model's predictive capabilities, shedding light on the physicochemical properties crucial for small molecule accumulation. Our findings highlight the importance of considering statistical moments beyond mean values in predictive modeling and suggest avenues for future research. Overall, our study provides insights into the complex dynamics of antibiotic accumulation in Escherichia coli bacterial cells, generalizable to other Gram-negative species, offering a promising approach for the discovery of effective antibacterial agents, identifying new hits, and improving them to define effective lead agents.

Advancing Forest-Fire Management: Exploring Sensor Networks, Data Mining Techniques, and SVM Algorithm for Prediction

Forest-fire is a pressing global problem that has far-reaching effects on human life and the environment, with climate change exacerbating their frequency and intensity. There is an urgent need for advanced predictive systems to mitigate these impacts. To address this issue, this study introduces a forest-fire prediction framework integrating wireless sensor networks (WSNs), data analysis, and machine learning. Sensor nodes deployed in a forest area collected real-time meteorological data, which was transmitted using LoRaWAN technology. Data mining techniques prepared the data for analysis using the SVM algorithm, revealing relationships between meteorological parameters and wildfire risk. The SVM model demonstrated an accuracy of 86% in classifying forest-fire risk levels based on temperature, humidity, wind speed, and rainfall data. The integrated framework of WSNs and the SVM algorithm provides a high-accuracy model for forest-fire risk prediction. The model is compared to the Canadian Forest Fire Hazard Rating System to validate its accuracy, demonstrating strong agreement with historical records and reports. The model's practical implications include efficient management, early detection, and prevention strategies. However, the model's limitations suggest avenues for future research, we should consider broader geographic applications and using advanced machine-learning methods to enhance the model's predictive capabilities.

Evaluation of Total Risk-Weighted Assets in Islamic Banking through Fintech Innovations

The assessment of total risk-weighted assets (LTRWAs) in the banking sector is of the utmost importance. It serves as a critical component for regulatory compliance, risk management, and capital adequacy. By accurately assessing LTRWAs, banks can effectively meet regulatory requirements, efficiently allocate capital resources, and proactively manage risks. Moreover, the accurate assessment of LTRWAs supports performance evaluation and fosters investor confidence in the financial stability of banks. This study presents statistical analyses and machine learning methods to identify factors influencing LTRWAs. Data from Bahrain, Jordan, Qatar, the United Arab Emirates, and Yemen, spanning from 2010 to 2021, was utilized. Various statistical tests and models, including ordinary least squares, fixed effect, random effect, correlation, variance inflation factor, tolerance tests, and fintech models, were conducted. The results indicated significant impacts of the unemployment rate, inflation rate, natural logarithm of the loan-to-asset ratio, and natural logarithm of total assets on LTRWAs in regression models. The dataset was divided into a training group (90% of the data) and a testing group (10% of the data) to evaluate the predictive capabilities of various fintech models, including an adaptive network-based fuzzy inference system (ANFIS), a hybrid neural fuzzy inference system (HyFIS), a fuzzy system with the heuristic gradient descent (FS.HGD), and fuzzy inference rules with the descent method (FIR.DM) models. The selection of the optimal model is contingent upon assessing its performance according to specific error criteria. The HyFIS model outperformed others with lower errors in predicting LTRWAs. Independent t-tests confirmed statistically significant differences between original and predicted LTRWA for all models, with HyFIS showing closer predictions. This study provides valuable insights into LTRWA prediction using advanced statistical and machine learning techniques, based on a dataset from multiple countries and years.

How Effective are Tabletop Role-Playing (Serious) Games in Understanding and Validating the Predictive Capabilities of Disaster Response Agent-based Models?

How Effective are Tabletop Role-Playing (Serious) Games in Understanding and Validating the Predictive Capabilities of Disaster Response Agent-based Models?

Leveraging Artificial Neural Networks for Robust Landslide Susceptibility Mapping: A Geospatial Modeling Approach in the Ecologically Sensitive Nilgiri District, Tamil Nadu

Landslides pose a significant threat to the lives and livelihoods of marginalised communities residing in rural areas and the delicate ecological balance of the environment. Implementing advanced technologies is crucial for improving hazard risk assessment and enhancing preparedness measures in regions characterised by diverse topography and complex geological formations. Geospatial applications and modelling techniques have emerged as indispensable in mitigating landslide risks, particularly in environmentally sensitive areas. This study presents a comprehensive approach to landslide susceptibility mapping in the Nilgiri District of Tamil Nadu, India, leveraging the power of Artificial Neural Networks (ANNs) and integrating multi-dimensional geospatial datasets. Integrating ANN-based modelling and geospatial techniques offers significant advantages in terms of statistical robustness, reproducibility, and the ability to analyse the complex interplay of factors influencing landslide hazards quantitatively. The methodology involves rigorous pre-processing and integrating spatial data, including landslide event occurrences as the dependent variable and ten independent parameters influencing landslide susceptibility. These parameters encompass elevation, slope aspect, slope degree, distance to roads, land use patterns, geomorphology, lithology, drainage density, lineament density, and rainfall distribution. Feature extraction and selection techniques are employed to effectively model the complex interactions between these factors and landslide occurrences. This process identifies the most relevant variables influencing landslide susceptibility, enhancing the model's predictive capabilities. The state-of-the-art ANNs are trained using historical landslide occurrence data and the selected influencing factors, enabling the development of a robust and accurate landslide susceptibility model. The performance of the developed model is rigorously evaluated using a comprehensive suite of metrics, including accuracy, precision, and the Area under the Receiver Operating Characteristic (ROC) curve. Preliminary results indicate that the ANN-based landslide susceptibility model outperforms traditional zonation methods, demonstrating higher accuracy and reliability in predicting landslide-prone areas. The resulting Landslide Susceptibility Map (LSM) categorises the study area into five distinct hazard zones, ranging from very high (664.1 km2), high (598.9 km2), moderate (639.7 km2), low (478.9 km2) and to very low (170.9 km2) . Notably, the eastern and central regions of the district emerge as particularly vulnerable to landslide occurrences. The study's findings have far-reaching implications for disaster risk reduction efforts, land-use planning, and sustainable development strategies in the ecologically sensitive Nilgiri District and beyond.

Evaluation of the prediction capability of mesoscale CFD models on velocity fluctuations in liquid-solid flows

Mesoscale CFD models are of great importance in multiphase flow studies due to their acceptable computational cost and great numerical fidelity. However, the evaluations of mesoscale models on velocity fluctuation predictions are still limited. An evaluation of the predictive capability of current CFD models on velocity fluctuations in liquid-solid flows is presented. Numerical simulations are carried out with both the Eulerian-Eulerian Two-Fluid Model (TFM) and the Eulerian-Lagrangian Discrete Phase Model (DPM), and compared with the available experimental data. Together with the proper models in both phases, the TFM satisfactorily predicts two-phase velocity fluctuations. A diffusion-based coupling method is implemented to mitigate the grid dependency problem in the DPM. The diffusion-based coupling method greatly improves the predictive fidelity of the particle volume fraction for DPM. However, it overpredicts velocity fluctuations for both phases. Further research is required to enhance the fidelity of velocity fluctuation prediction in fluid-particle flow simulation models.

Can Graph Machines Accurately Estimate 13C NMR Chemical Shifts of Benzenic Compounds?

In the organic laboratory, the 13C nuclear magnetic resonance (NMR) spectrum of a newly synthesized compound remains an essential step in elucidating its structure. For the chemist, the interpretation of such a spectrum, which is a set of chemical-shift values, is made easier if he/she has a tool capable of predicting with sufficient accuracy the carbon-shift values from the structure he/she intends to prepare. As there are few open-source methods for accurately estimating this property, we applied our graph-machine approach to build models capable of predicting the chemical shifts of carbons. For this study, we focused on benzene compounds, building an optimized model derived from training a database of 10,577 chemical shifts originating from 2026 structures that contain up to ten types of non-carbon atoms, namely H, O, N, S, P, Si, and halogens. It provides a training root-mean-squared relative error (RMSRE) of 0.5%, i.e., a root-mean-squared error (RMSE) of 0.6 ppm, and a mean absolute error (MAE) of 0.4 ppm for estimating the chemical shifts of the 10k carbons. The predictive capability of the graph-machine model is also compared with that of three commercial packages on a dataset of 171 original benzenic structures (1012 chemical shifts). The graph-machine model proves to be very efficient in predicting chemical shifts, with an RMSE of 0.9 ppm, and compares favorably with the RMSEs of 3.4, 1.8, and 1.9 ppm computed with the ChemDraw v. 23.1.1.3, ACD v. 11.01, and MestReNova v. 15.0.1-35756 packages respectively. Finally, a Docker-based tool is proposed to predict the carbon chemical shifts of benzenic compounds solely from their SMILES codes.

Analysis of TEC variations and prediction of TEC by RNN during Indonesian earthquakes occurred from 2004 to 2024 and comparison with IRI-2020 model

The Ionosphere, a crucial region of Earth’s atmosphere extending from approximately 50 to 1000 km above the Earth’s surface, plays a pivotal role in global communication, navigation, and satellite-based technologies. Among its various parameters, Total Electron Content (TEC) stands out as a key metric, representing the integrated electron density along a specific path through the ionosphere. This research focuses on analyzing TEC variations during Indonesian earthquakes between 2004 and 2024 using a Recurrent Neural Network (RNN) model. The study investigates the ionospheric response to seismic activity, particularly in Indonesia, a region prone to earthquakes due to its location within the Pacific Ring of Fire and complex tectonic setting involving multiple lithospheric plates. The RNN model, designed to predict TEC variations during earthquakes, utilizes data on solar and geomagnetic activity, including the Kp index, solar wind, and geomagnetic storm activity, obtained from the OMNIWEB data centre, along with TEC data from the IONOLAB data servers for the BAKO station in Cibinong, Indonesia. The earthquakes considered for observation include significant events such as the 2004 Indian Ocean earthquake and tsunami, the 2012 Wharton Basin earthquake, and subsequent seismic events up to 2024. Statistical metrics such as Mean Bias Deviation (MBD), Relative Error (REL_E), Absolute Error (ABS_E), and Root Mean Square Error (RMSE) are used to evaluate the predictive capability of the RNN model and compare it with the International Reference Ionosphere (IRI) 2020 model, a widely accepted empirical model for describing ionospheric parameters. The comparison reveals the performance of the RNN model in capturing TEC variations during seismic events, providing valuable insights for earthquake monitoring and early warning systems.

Comparative analysis of the predictive capabilities of some machine learning models: A case study using wind speed data

The widespread use of fossil fuels for global energy production significantly contributes to global warming. This study presents a comparative analysis of various machine learning models, which are the long short-term memory (LSTM) network, support vector regression (SVR), and gradient boosting method (GBM). Gaussian process regression (GPR) is a benchmark model across different forecasting horizons. The study uses South African wind speed data from 1 January 2018 to 31 December 2021, sourced from the Western Cape province. The dataset underwent preprocessing, and diverse feature selection techniques were implemented to enhance model accuracy. Performance evaluation of the models was done using mean absolute error (MAE), root mean squared error (RMSE), and mean absolute scaled error (MASE). Results indicate that SVR exhibits superior accuracy to other models for two distinct forecast horizons (h = 670 and h = 1339), respectively. Additionally, GPR surpasses other models for the forecasting horizon h = 224. This study provides insights into the comparative strengths and weaknesses of different machine learning models for wind speed prediction, which could be useful in selecting an appropriate model for future applications in renewable energy and weather forecasting. Potential areas for future research include improving prediction accuracy via ensemble deep learning algorithms and incorporating additional meteorological variables. Moreover, investigating temporal dynamics, broadening geographical coverage and integrating uncertainty quantification methods can improve wind speed prediction, thereby facilitating more effective renewable energy planning and decision-making processes

Research on stamping formability and process simulation of AZ31 magnesium alloy based on deep learning

To study the formability of AZ31 magnesium alloy under different temperature conditions and improve the forming quality of its parts, this paper aims to solve the minimization problem of pursuing the “maximum thinning rate”. This study obtained the mechanical properties of AZ31 magnesium alloy at 25 °C, 150 °C, 250 °C, and 350 °C through uniaxial tensile tests and explored the stamping formability of AZ31 magnesium alloy mobile phone cases using the finite element inverse calculation method (MSTEP). Based on the hot stamping of AZ31 magnesium alloy at 250 °C, the Monte Carlo Simulation (MCS) method was used to randomly sample the blank holder force, stamping speed, friction coefficient, and resistance coefficient, and then simulate to obtain the corresponding “maximum thinning rate”. In addition, a hybrid prediction model optimized by Genetic Algorithm (GA) and Dung Beetle Optimization (DBO) based on the Back Propagation Neural Network (BPNN) - Random Forest (RF) was used to explore the nonlinear relationship between blank holder force, stamping speed, friction coefficient, resistance coefficient, and the “maximum thinning rate”. The results show that the plasticity of AZ31 magnesium alloy enhances with the temperature increase, especially exhibiting the best stamping formability at 250 °C. Feature importance indicates that the resistance coefficient and stamping speed have a greater impact on the maximum thinning rate, while the influence of blank holder force and friction coefficient is relatively smaller. Mean Square Error (MSE) shows that the predictive capability of the hybrid model is significantly superior to that of the single BP neural network and Random Forest model, displaying optimal efficiency. The predicted maximum thinning rate by the model is 15.39 %, with an error rate of only 0.32 % compared to the DYNAFORM finite element software simulation results, confirming the correctness, accuracy, and effectiveness of the prediction model. Overall, this study provides practical optimization strategies for metal stamping forming processes, with significant practical application value.

Comparison of mathematical and supervised machine-learning models for ductile-to-brittle transition in bcc alloys

This study focuses on modelling Ductile-to-Brittle Transition (DBT) curves using various mathematical and supervised machine learning models. Charpy impact energy values are converted to normalized energy values to account for reductions in upper-shelf energy. The research introduces a saturation parameter in mathematical models to capture these variations and examines the influence of alloying elements, microstructure, and neutron irradiation on DBT behaviour in nuclear structural materials. Detailed analyses reveal how fitting parameters vary with these factors and demonstrate that mathematical models’ fitting parameters generally align with observed DBT curve trends. The predictive capabilities of these mathematical models are also compared with those of supervised machine learning models, highlighting the strengths and limitations of each approach in modelling DBT behaviour. An explainable approach is used for interpretation of machine learning models and it is shown that this approach can be effectively used for the influence of various independent parameters on impact energy.

A temporal prediction model for ship maneuvering motion based on multi-head attention mechanism

In real-world maritime conditions, accurately predicting ship maneuvering motion over ultrashort periods can enhance the development of more precise vessel control algorithms. This study proposes a prediction model for ship maneuvering that utilizes a multi-head attention mechanism. To form a basis for this data-driven approach, we gathered training data from turning and zigzag tests of a ship conducted in natural sea environments. The model's predictive capability is assessed by comparing two strategies over a fixed 30-time-step forecast horizon: iterative multi-step forecasting (IMS) and direct multi-step forecasting (DMS). Notably, the model employs distinct attention layers to examine the temporal relationships among various motion variables across the three degrees of freedom in the output. Moreover, it uses separate attention layers to explore the distinct time response traits of the input motion states and control signals. The results demonstrate that different motion state variables display unique temporal correlations under different signal sequences, and the proposed network architecture effectively captures these temporal response traits.

Agent-based model demonstrates the impact of nonlinear, complex interactions between cytokinces on muscle regeneration.

Muscle regeneration is a complex process due to dynamic and multiscale biochemical and cellular interactions, making it difficult to identify microenvironmental conditions that are beneficial to muscle recovery from injury using experimental approaches alone. To understand the degree to which individual cellular behaviors impact endogenous mechanisms of muscle recovery, we developed an agent-based model (ABM) using the Cellular-Potts framework to simulate the dynamic microenvironment of a cross-section of murine skeletal muscle tissue. We referenced more than 100 published studies to define over 100 parameters and rules that dictate the behavior of muscle fibers, satellite stem cells (SSCs), fibroblasts, neutrophils, macrophages, microvessels, and lymphatic vessels, as well as their interactions with each other and the microenvironment. We utilized parameter density estimation to calibrate the model to temporal biological datasets describing cross-sectional area (CSA) recovery, SSC, and fibroblast cell counts at multiple timepoints following injury. The calibrated model was validated by comparison of other model outputs (macrophage, neutrophil, and capillaries counts) to experimental observations. Predictions for eight model perturbations that varied cell or cytokine input conditions were compared to published experimental studies to validate model predictive capabilities. We used Latin hypercube sampling and partial rank correlation coefficient to identify in silico perturbations of cytokine diffusion coefficients and decay rates to enhance CSA recovery. This analysis suggests that combined alterations of specific cytokine decay and diffusion parameters result in greater fibroblast and SSC proliferation compared to individual perturbations with a 13% increase in CSA recovery compared to unaltered regeneration at 28 days. These results enable guided development of therapeutic strategies that similarly alter muscle physiology (i.e. converting extracellular matrix [ECM]-bound cytokines into freely diffusible forms as studied in cancer therapeutics or delivery of exogenous cytokines) during regeneration to enhance muscle recovery after injury.