Predictive Capability Of Model Research Articles

A Bayesian Measure of Model Accuracy.

Ensuring that the proposed probabilistic model accurately represents the problem is a critical step in statistical modeling, as choosing a poorly fitting model can have significant repercussions on the decision-making process. The primary objective of statistical modeling often revolves around predicting new observations, highlighting the importance of assessing the model's accuracy. However, current methods for evaluating predictive ability typically involve model comparison, which may not guarantee a good model selection. This work presents an accuracy measure designed for evaluating a model's predictive capability. This measure, which is straightforward and easy to understand, includes a decision criterion for model rejection. The development of this proposal adopts a Bayesian perspective of inference, elucidating the underlying concepts and outlining the necessary procedures for application. To illustrate its utility, the proposed methodology was applied to real-world data, facilitating an assessment of its practicality in real-world scenarios.

GM(1,N) method for the prediction of critical failure pressure of type III tank in fire scenarios

The determination of the pressure-bearing performance of hydrogen storage tanks (HST) is integral to enhancing their operational safety and risk management capabilities. In this study, an inaugural development of low-cost critical failure pressure (BPc) prediction model for small samples was proposed, which was based on a combination of the fuzzy grey relational analysis (FGRA) and GM(1,N). Specifically, a total of 15 of BPc-related factors were proposed for analysis and projections based on the data from bonfire tests. Results from the FGRA indicated burst pressure was most closely linked to the initial filling pressure (Rij1 = 0.939) of tanks under fire scenario. The normal working pressure (Rij2 = 0.924) and the elasticity modulus (Rij3 = 0.871) also demonstrating significant impacts. Furthermore, three GM(1,N) prediction models which could estimate BPc with multi-factor coupling were developed. Increasing the number of input feature factors could enhance the predictive capability of GM model. The GM(1,16) had optimal prediction performance among the models, achieving a prediction accuracy of 99.8 %. As well, the mean absolute error (MAE) was 0.799 MPa while the mean absolute percentage error (MAPE) was 1.56 %. This paper offered a novel option for establishing the HST critical failure criterion safely and efficiently.

Assessing the predictive capability of machine learning models in determining clinical outcomes for patients with cervical spondylotic myelopathy treated with laminectomy and posterior spinal fusion

BackgroundCervical spondylotic myelopathy (CSM) is a prevalent degenerative condition resulting from spinal cord compression and injury. Laminectomy with posterior spinal fusion (LPSF) is a commonly employed treatment approach for CSM patients. This study aimed to assess the effectiveness of machine learning models (MLMs) in predicting clinical outcomes in CSM patients undergoing LPSF.MethodsA retrospective analysis was conducted on 329 CSM patients who underwent LPSF at our institution from Jul 2017 to Jul 2023. Neurological outcomes were evaluated using the modified Japanese Orthopaedic Association (mJOA) scale preoperatively and at the final follow-up. Patients were categorized into two groups based on clinical outcomes: the favorable group (recovery rates ≥ 52.8%) and the unfavorable group (recovery rates < 52.8%). Potential predictors for poor clinical outcomes were compared between the groups. Four MLMs—random forest (RF), logistic regression (LR), support vector machine (SVM), and k-nearest neighborhood (k-NN)—were utilized to predict clinical outcome. RF model was also employed to identify factors associated with poor clinical outcome.ResultsOut of the 329 patients, 185 were male (56.2%) and 144 were female (43.4%), with an average follow-up period of 17.86 ± 1.74 months. Among them, 267 patients (81.2%) had favorable clinical outcomes, while 62 patients (18.8%) did not achieve favorable results. Analysis using binary logistic regression indicated that age, preoperative mJOA scale, and symptom duration (p < 0.05) were independent predictors of unfavorable clinical outcomes. All models performed satisfactorily, with RF achieving the highest accuracy of 0.922. RF also displayed superior sensitivity and specificity (sensitivity = 0.851, specificity = 0.944). The Area under the Curve (AUC) values for RF, Logistic LR, SVM, and k-NN were 0.905, 0.827, 0.851, and 0.883, respectively. The RF model identified preoperative mJOA scale, age, symptom duration, and MRI signal changes as the most significant variables associated with poor clinical outcomes in descending order.ConclusionsThis study highlighted the effectiveness of machine learning models in predicting the clinical outcomes of CSM patients undergoing LPSF. These models have the potential to forecast clinical outcomes in this patient population, providing valuable prognostic insights for preoperative counseling and postoperative management.

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.

Agent-based model demonstrates the impact of nonlinear, complex interactions between cytokines 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.

Anomaly Detection Using Machine Learning in Hydrochemical Data From Hot Springs: Implications for Earthquake Prediction

AbstractThis study explores the potential of machine learning algorithms for earthquake prediction, utilizing fluid chemical anomaly data from hot springs. Six hot springs, located within an active fault zone along the southeastern coast of China, were carefully chosen as hydrochemical monitoring sites for an extended period of two and a half years. Using this data, a prediction model integrating six algorithms was developed to forecast M ≥ 5 earthquakes in Taiwan. The model's performance was validated against recorded earthquake events, and the factors influencing its predictive capability were analyzed. Our comprehensive analysis conclusively demonstrates the superiority of machine learning algorithms over traditional statistical methods for earthquake prediction. Additionally, including sampling time in the data sets significantly improves the model's predictive performance. However, it is important to note that the model's predictive performance varies across different hot spring and indicators type, highlighting the importance of identifying optimal indicators for specific scenarios. The model parameters, including the anomaly detection rate (P) and earthquake response time threshold (M), significantly impact the model's predictive capabilities. Therefore, adjustments are needed to optimize the model's performance for practical use. Despite limitations such as the inability to differentiate pre‐earthquake anomalies from post‐earthquake anomalies and pinpoint the precise location of earthquakes, this study successfully showcases the potential of machine learning algorithms in earthquake prediction, paving the way for further research and improved prediction methods.

Symbolic regression as a feature engineering method for machine and deep learning regression tasks

In the realm of machine and deep learning (DL) regression tasks, the role of effective feature engineering (FE) is pivotal in enhancing model performance. Traditional approaches of FE often rely on domain expertise to manually design features for machine learning (ML) models. In the context of DL models, the FE is embedded in the neural network’s architecture, making it hard for interpretation. In this study, we propose to integrate symbolic regression (SR) as an FE process before a ML model to improve its performance. We show, through extensive experimentation on synthetic and 21 real-world datasets, that the incorporation of SR-derived features significantly enhances the predictive capabilities of both machine and DL regression models with 34%–86% root mean square error (RMSE) improvement in synthetic datasets and 4%–11.5% improvement in real-world datasets. In an additional realistic use case, we show the proposed method improves the ML performance in predicting superconducting critical temperatures based on Eliashberg theory by more than 20% in terms of RMSE. These results outline the potential of SR as an FE component in data-driven models, improving them in terms of performance and interpretability.

Investigating the Impact of Data Normalization Methods on Predicting Electricity Consumption in a Building Using different Artificial Neural Network Models.

The study investigates the impact of data normalization on the prediction of electricity consumption in buildings using four multilayer Artificial Neural Networks (ANN) algorithms: Long Short-Term Memory Networks (LSTM), Levenberg-Marquardt Back-propagation (LMBP), Recurrent Neural Networks (RNN), and General Regression Neural Network (GRNN). Four data normalization approaches, Min-Max Scaling, Mean, Z-score, and Gaussian function were assessed on experimental datasets. The LSTM algorithm, when combined with Min-Max normalization, showed the most favorable predictive capabilities, with a low Coefficient of Variation of the Root Mean Square Error (CVRMSE) of 10.3 and Normalized Mean Bias Error (NMBE) of 0.6. The remaining three normalization approaches showed satisfactory concordance with empirical data, but with slight disparities in precision. The LMBP model, when using Z-score normalization, had favorable performance in forecasting electricity consumption, but the discrepancies across the models were not significant. The Recurrent Neural Network (RNN) model, when used with Gaussian normalization, exhibited the most favorable performance, with the lowest Coefficient of Variation of Root Mean Square Error (CVRMSE) at 11.8 and Normalized Mean Biased Error (NMBE) at 0.6. The Generalized Regression Neural Network (GRNN) model, trained on unprocessed data, exhibited superior performance, with the lowest Coefficient of Variation of Root Mean Square Error (CVRMSE) at 19.2 and NMBE at 1.0. In conclusion, the study highlights the significant influence of data normalization on the predictive capabilities of various ANN models, suggesting that careful use of data normalization techniques can significantly improve the accuracy of electricity consumption forecasting in buildings.

Immune-related adverse event prediction and influential factor identification.

e14650 Background: Immune checkpoint inhibitors (ICIs), have demonstrated efficacy in treating patients with non-small cell lung cancer (NSCLC) and advanced melanoma. However, ICIs can lead to immune-related adverse events (irAE), which will cause discontinuance of the ICI treatment. The objective of this study is to identify factors that predict which patients with NSCLC or melanoma will develop irAE(s) from ICIs. Methods: We curated the patient cohort from the Immune-Oncology data repository developed at the Georgetown-Lombardi Comprehensive Cancer Center. This cohort encompassed 466 patients (245 individuals with NSCLC and 221 with advanced melanoma) who received single-agent anti-PD-(L)1 or combination of anti-PD-1 and anti-CTLA4 therapy. Parameters selected for prediction comprises baseline demographics, laboratory results, treatment information, and cancer-related mutations. The classification of irAE were identified by CTCAE V4.03 guidelines. Overall, 219 patients experienced one or more irAE, while the remaining 247 patients were irAE-free. We employed machine learning models for prediction, which were trained using 80% of the dataset with a five-fold cross-validation, and their performance was evaluated using the area under the receiver-operating curve (AUROC) and F1 score. Furthermore, we identified the key factors that significantly impacted the model by SHAP values. Results: Nine machine learning models were deployed, including logistic regression, support vector machine, gradient boosting (GB), etc. The GB model demonstrated the highest performance, achieving an AUROC score of 75% and an F1 score of 77%. According to the SHAP value obtained from the GB model, NSCLC patients exhibited a higher propensity for irAE development compared to melanoma patients. The top three features associated with a heightened likelihood of irAE development are the treatment of the combination of anti PD-1 and anti CTLA-4, a higher albumin/globulin ratio, and a pre-treatment ECOG Performance Status score of 0. The top three features associated with a lower likelihood of irAE development are an elevated red cell distribution width, the treatment with single agent anti-PD-1, and a higher globulin level. Conclusions: Our research harnesses a comprehensive set of variables to forecast the development of irAEs in NSCLC and melanoma patients undergoing ICI treatment. The results underscore the predictive capabilities of ML models. Investigating the connections between pivotal biomarkers, such as globulin and red cell distribution width, has the potential to offer insights that enable patients to mitigate irAE. These findings emphasize the promising potential of ML techniques not only in predicting the likelihood of irAE but also in the grade of irAE. To mitigate the limitation of a relatively small sample size, we are actively working to expand the data registry by collaborating with additional institutions.

Feasibility of Formulating Ecosystem Biogeochemical Models From Established Physical Rules

AbstractTo improve the predictive capability of ecosystem biogeochemical models (EBMs), we discuss the feasibility of formulating biogeochemical processes using physical rules that have underpinned the many successes in computational physics and chemistry. We argue that the currently popular empirically based approaches, such as multiplicative empirical response functions and the law of the minimum, will not lead to EBM formulations that can be continuously refined to incorporate improved mechanistic understanding and empirical observations of biogeochemical processes. Instead, we propose that EBM parameterizations, as a lossy data compression problem, can be better formulated using established physical rules widely used in computational physics and chemistry, and different biogeochemical processes can be more robustly integrated within a reactive‐transport framework. Through several examples, we demonstrate how mathematical representations derived from physical rules can improve understanding of relevant biogeochemical processes and enable more effective communication between modelers, observationalists, and experimentalists regarding essential questions, such as what measurements are needed to meaningfully inform models and how can models generate new process‐level hypotheses to test in empirical studies. Finally, while empirical models with more parameters are often less robust, physical rules‐based models can be more robust and show lower predictive equifinality, stemming from their enhanced consistency in representations of processes, interactions and spatial scaling.

Dialearner: Predicting Diabetes with Machine Learning

Abstract: In the medical field, it is essential to predict diseases early to prevent them. Diabetes is one of the mostdangerous diseases all over the world. In modern lifestyles, sugar and fat are typically present in our dietary habits, which have increased the risk of diabetes. To predict the disease, it is extremely important to understandits symptoms. Currently, machine-learning (ML) algorithms are valuable for disease detection. Diabetes, in all its types, costs countries of all income levels unacceptably enormous personal, societal, and economic expenses.The proposed system can help doctors to make data-driven decisions and enhance patients’ treatment. Several machine learning algorithms that are Decision Tree, Support Vector Machine, Random Forest, Artificial NeuralNetwork, k-Nearest Neighbors, Logistic Regression, and Naive Bayes are used. Evaluation metrics such as accuracy, precision, recall, and F1-score are utilized to assess the model's predictive capability. Cross-validationtechniques are employed to ensure robustness and generalizability. The proposed model holds significant promisein facilitating early detection and intervention for individuals at risk of developing diabetes, thereby improving patient outcomes, and reducing healthcare burden. Future research directions may include incorporating additional features and exploring ensemble learning techniques to further enhance predictive accuracy and reliability

An improved convolutional neural network for predicting porous media permeability from rock thin sections

Permeability prediction is a crucial aspect of investigating fluid flow capacity in porous media, and rapidly predicting the permeability of rock thin sections aids in reservoir assessment. In recent years, significant progress has been made regarding the use of deep learning for predicting porous media permeability, effectively overcoming the limitations associated with time-consuming conventional permeability determination methods. To address the issues of low prediction accuracy and insufficient generalization capability of previous permeability prediction models, this study proposes an improved convolutional neural network for porous media permeability and applies it to predict the permeability of rock thin sections. Given the difficulty of obtaining a large amount of rock thin section data, we synthetically generated 1000 binary images of porous media and calculated the corresponding permeability labels using computational fluid dynamics. These data were used as foundational data for training the model. The proposed model achieved accurate and efficient permeability prediction of porous media with a correlation of 0.9667 on testing set. Simultaneously, we employed class activation mapping technology to visualize the permeability prediction process of the neural network, thereby elucidating the working principles of the neural network during porous media permeability prediction. Subsequently, based on an optimally pretrained neural network, transfer learning was utilized to predict the permeability of rock thin sections, which achieved a correlation of 0.9032. Our study extends the rock thin section applications and provides technical support for reservoir evaluation.

Spatial analysis and machine learning: Towards integrated predictive modeling advancements

This paper presents a comprehensive exploration of the integration of spatial analysis with machine learning techniques, aiming to enhance predictive modeling capabilities across various domains. Spatial analysis, a methodological approach for understanding geographic patterns and relationships, when combined with the computational power of machine learning, offers unprecedented opportunities for analyzing complex spatial datasets. Through quantitative analysis and the application of mathematical models, this study demonstrates the effectiveness of this integration in improving the accuracy and efficiency of predictive models. The research encompasses a range of applications, from environmental monitoring to urban planning and public health, showcasing the versatility and potential of combining spatial analysis with machine learning.

Spatiotemporal Mechanism-Based Spacetimeformer Network for InSAR Deformation Prediction and Identification of Retrogressive Thaw Slumps in the Chumar River Basin

The increasing incidence of retrogressive thaw slumps (RTSs) in permafrost regions underscores the need for detailed spatial and temporal analysis using InSAR technology to monitor and predict dynamic changes in the process of RTSs. Nevertheless, current InSAR deformation forecasting methods employing deep learning strategies such as the traditional long short-term memory (LSTM) and recent transformer models encounter difficulties in effectively capturing temporal features. Moreover, they are limited in their ability to directly integrate spatial information. In this paper, an innovative deep learning approach named Spacetimeformer is proposed for predicting medium- and short-term InSAR deformation of RTSs in the Chumar River area. This method employs a transformer architecture with a spatiotemporal attention mechanism, which enhances the long-term prediction capabilities of time series models and dynamic spatial modeling. It is applicable to multivariate InSAR spatiotemporal deformation prediction problems. The findings include a list of 72 RTSs compiled based on derived InSAR deformation maps and Sentinel-2 optical images, of which 64 have an average deformation rate exceeding 10 mm/year, indicating signs of permafrost degradation. The density distribution of the displacement maps predicted by the Spacetimeformer model aligned well with the InSAR deformation maps obtained from the small baseline subset (SBAS) method, with the overall prediction deviation controlled within 20 mm. In addition, the point-scale prediction results were compared with LSTM and transformer models. This study indicates that the Spacetimeformer network achieved good results in predicting the deformation of RTSs, with a root mean square error of 1.249 mm. The Spacetimeformer method for deformation prediction with the spacetime mechanism presented in this study can serve as a general framework for multivariate deformation prediction based on InSAR results. It can also quantitatively assess the spatial deformation characteristics and deformation trends of RTSs.

APEX-pHLA: A novel method for accurate prediction of the binding between exogenous short peptides and HLA class I molecules

Human leukocyte antigen (HLA) molecules play critically significant role within the realm of immunotherapy due to their capacities to recognize and bind exogenous antigens such as peptides, subsequently delivering them to immune cells. Predicting the binding between peptides and HLA molecules (pHLA) can expedite the screening of immunogenic peptides and facilitate vaccine design. However, traditional experimental methods are time-consuming and inefficient. In this study, an efficient method based on deep learning was developed for predicting peptide-HLA binding, which treated peptide sequences as linguistic entities. It combined the architectures of textCNN and BiLSTM to create a deep neural network model called APEX-pHLA. This model operated without limitations related to HLA class I allele variants and peptide segment lengths, enabling efficient encoding of sequence features for both HLA and peptide segments. On the independent test set, the model achieved Accuracy, ROC_AUC, F1, and MCC is 0.9449, 0.9850, 0.9453, and 0.8899, respectively. Similarly, on an external test set, the results were 0.9803, 0.9574, 0.8835, and 0.7863, respectively. These findings outperformed fifteen methods previously reported in the literature. The accurate prediction capability of the APEX-pHLA model in peptide-HLA binding might provide valuable insights for future HLA vaccine design.

A prognostic binary classifier comprised of five critical mRNAs stratified pancreatic cancer patients following resection

BackgroundPancreatic cancer is characterized by an extremely poor prognosis, even following potentially curative resection. Classical prognostic markers such as histopathological or clinical parameters have limited predictive power. The present study aimed to establish a prognostic model combining mRNA expression data with histopathological and clinical data to better predict survival and stratify pancreatic cancer patients following resection. We pioneered three models in one study and systematically evaluated the clinical benefits of all three models. MethodsTo identify differentially expressed genes in pancreatic cancer, mRNA data from normal (GTEx database) and pancreatic cancer (TCGA database) tissues were used. Survival analysis was carried out to identify prognosis-relevant genes from the identified differentially expressed genes and LASSO regression was used to filter out hub genes. The risk score of several hub genes was calculated according to gene expression and coefficients. Validation was carried out using an independent set of GEO microarray data. Multivariate COX regression was used for identifying independent clinical and pathological risk factors related to patient's survival in the TCGA database and a prognostic model combining mRNA expression data with histopathological and clinical data was established. Another prognostic model using clinicopathological factors from the SEER database was conceived based on multivariate COX regression. NRI (net reclassification improvement) and IDI (integrated discrimination index) were used to compare the predictive capabilities of the different models. ResultsWe identified 1589 differentially expressed genes (DEGs) through the comparison of normal and pancreatic cancer tissues, of whom 317 were associated with prognosis(p < 0.05). LASSO regression identified five hub genes, MYEOV, ANXA2P2, MET, CEP55, and KRT7, that were used for the five-mRNA-classifier prognostic model. The classifier could stratify patients into a short and long survival group: 5-year overall survival in the training set (TCGA, 6 % vs 52 %, p < 0.001), test set (TCGA, 18 % vs 55 %,p < 0.01) and external validation set (GEO, 0 % vs 25 %, p < 0.05). Sensitivity analysis showed that the mRNA model (model 1) was better than the clinicopathological no-mRNA model (model 2) in predicting 5-year survival in the TCGA database (AUC: 0.877 vs 0.718, z = 3.165, p < 0.01) and better than the multi-factor prognostic model (model 3) from the SEER database (AUC: 0.754, z = 2.637, p < 0.01). On predictive performance, model 1 improved model 2 (NRI = 0.084, z = 1.288, p = 0.198; IDI = 0.055, z = 1.041,p = 0.298) and model 3 (NRI = 0.167,z = 1.961,p = 0.05; IDI = 0.086, z = 1.427, p = 0.154). ConclusionThe five-mRNA-classifier is a reliable and feasible instrument to predict the prognosis of pancreatic cancer patients following resection. It might help in patiens counseling and assist clinicians in providing individualized treatment for patients in different risk groups.

Spatiotemporal Variation, Meteorological Driving Factors, and Statistical Models Study of Lake Surface Area in the Yellow River Basin

The surface area changes of 151 natural lakes over 37 months in the Yellow River Basin, based on remote sensing data and 21 meteorological indicators, employing spatial distribution feature analysis, principal component analysis (PCA), correlation analysis, and multiple regression analysis, identify key meteorological factors influencing these variations and their interrelationships. During the study period, lake area averages were from 0.009 km2 to 506.497 km2, with standard deviations ranging from 0.003 km2 to 184.372 km2. The coefficient of variation spans from 3.043 to 217.436, indicating considerable variability in lake area stability. Six primary meteorological factors were determined to have a significant impact on lake surface area fluctuations: 24 h precipitation, maximum daily precipitation, hours of sunshine, maximum wind speed, minimum relative humidity, and lakes in the source region of the Yellow River generally showed a significant positive correlation. For maximum wind speed (m/s), 28 lakes showed significant correlations, with five positive and twenty-three negative correlations, correlation coefficients ranging from −0.34 to −0.63, average −0.47, indicating an overall negative correlation between lake surface area and maximum wind speed. For maximum daily precipitation (mm), 36 lakes had 21 showing a positive correlation, indicating a positive correlation between lake surface area and daily precipitation in larger lakes. Furthermore, of the 117 lakes with sufficient data to model, the predictive capabilities of various models for lake surface area changes showcased distinct advantages, with the random forest model outperforming others in a dataset of 65 lakes, Ridge regression is best for 28 lakes, Lasso regression performs best for 20 lakes, Linear model is only best for 4 cases. The random forest model provides the best fit due to its ability to handle a large number of feature variables and consider their interactions, thereby offering the best fitting effect. These insights are crucial for understanding the influence of meteorological factors on lake surface area changes within the Yellow River Basin and are instrumental in developing predictive models based on meteorological data.

Prediction and optimisation of gasoline quality in petroleum refining: The use of machine learning model as a surrogate in optimisation framework

AbstractHardware‐based sensing frameworks such as cooperative fuel research engines are conventionally used to monitor research octane number (RON) in the petroleum refining industry. Machine learning techniques are employed to predict the RON of integrated naphtha reforming and isomerisation processes. A dynamic Aspen HYSYS model was used to generate data by introducing artificial uncertainties in the range of ±5% in process conditions, such as temperature, flow rates, etc. The generated data was used to train support vector machines (SVM), Gaussian process regression (GPR), artificial neural networks (ANN), regression trees (RT), and ensemble trees (ET). Hyperparameter tuning was performed to enhance the prediction capabilities of GPR, ANN, SVM, ET and RT models. Performance analysis of the models indicates that GPR, ANN, and SVM with R2 values of 0.99, 0.978, and 0.979 and RMSE values of 0.108, 0.262, and 0.258, respectively performed better than the remaining models and had the prediction capability to capture the RON dependence on predictor variables. ET and RT had an R2 value of 0.94 and 0.89, respectively. The GPR model was used as a surrogate model for fitness function evaluations in two optimisation frameworks based on genetic algorithm and particle swarm method. Optimal parameter values found by the optimisation methodology increased the RON value by 3.52%. The proposed methodology of surrogate‐based optimisation will provide a platform for plant‐level implementation to realise the concept of industry 4.0 in the refinery.

Enhanced predictive optimization of methane dry reforming via ResponseSurface methodology and artificial neural network approaches: Insights using a novel nickel-strontium-zirconium-aluminum catalyst

This study investigates the molecular dynamics of methane dry reforming catalyzed by a novel nickel-strontium-zirconium-aluminum (5Ni+3Sr/10Zr+Al) catalyst, leveraging both Response Surface Methodology (RSM) and Radial Basis Function Neural Network (RBFNN) for predictive optimization. Focusing on the impact of operational parameters—hourly space velocity, reaction temperature, and CO2:CH4 mole ratio—on the conversion rates and formation of reaction components, we aim to predict optimal conditions and corresponding process variables. Through a comparison of a three-layer Feed Forward Neural Network, optimized at a 3:10:1 topology, with traditional RSM approaches, our findings highlight the superior predictive capabilities of ANN models. Notably, ANN demonstrated exceptional performance with Radj2and F_Ratio values significantly surpass those of RSM, alongside lower statistical error terms. This superiority is attributed to ANN's robust handling of nonlinear relationships between inputs and outputs, asserting its potential for enhancing predictive accuracy in chemical process optimization. At optimum predicted conditions like 1 CH4/CO2,750 °C reaction temperature, 12000 cm3g−1h−1 space velocity, NiSrZrAl outperformed with > 85 % CH4 and CO2 conversion with H2/CO ∼1 up to 20 h time on stream. Our research underscores the importance of integrating advanced modeling techniques for the efficient and accurate prediction of catalytic reactions, offering valuable insights for future applications in chemical engineering and catalysis.

A Sensorized 3D-Printed Knee Test Rig for Preliminary Experimental Validation of Patellar Tracking and Contact Simulation.

Experimental validation of computational simulations is important because it provides empirical evidence to verify the accuracy and reliability of the simulated results. This validation ensures that the simulation accurately represents real-world phenomena, increasing confidence in the model's predictive capabilities and its applicability to practical scenarios. The use of musculoskeletal models in orthopedic surgery allows for objective prediction of postoperative function and optimization of results for each patient. To ensure that simulations are trustworthy and can be used for predictive purposes, comparing simulation results with experimental data is crucial. Although progress has been made in obtaining 3D bone geometry and estimating contact forces, validation of these predictions has been limited due to the lack of direct in vivo measurements and the economic and ethical constraints associated with available alternatives. In this study, an existing commercial surgical training station was transformed into a sensorized test bench to replicate a knee subject to a total knee replacement. The original knee inserts of the training station were replaced with personalized 3D-printed bones incorporating their corresponding implants, and multiple sensors with their respective supports were added. The recorded movement of the patella was used in combination with the forces recorded by the pressure sensor and the load cells, to validate the results obtained from the simulation, which was performed by means of a multibody dynamics formulation implemented in a custom-developed library. The utilization of 3D-printed models and sensors facilitated cost-effective and replicable experimental validation of computational simulations, thereby advancing orthopedic surgery while circumventing ethical concerns.