This paper examines the application of artificial intelligence models, specifically artificial neural networks (ANNs), to predict the peak particle velocity (PPV) at a tunnel lining resulting from nearby blasting. The ANN models are developed using data collected during the construction of the Hai Van 2 traffic tunnel, which connects Hue City and Da Nang City in Vietnam. The training and testing results above the training database and testing database (with a coefficient of determination R 2 of 0.9 913 and a mean square error MSE of 3.10 - 4 for the testing dataset, and R 2 = 0.9903 and MSE = 3.10 - 4 for the training dataset) demonstrate the model's high accuracy in predicting the peak particle velocity at the Hai Van 1 tunnel lining, relate to the impact of mine blasts during the construction of the Hai Van 2 tunnel.

In mining activities, surface subsidence is an extremely important issue. Therefore, predicting surface subsidence is a necessary task to ensure the safety and efficiency of mining operations. This study aims to evaluate the prediction results obtained from the Artificial Neural Networks (ANN) and Exponential Function (EF) methods for predicting surface subsidence in underground mining areas. To achieve this objective, data from four observation points on the Quang Hanh mine surface were utilized for comparison with the prediction results. For both methods, data from the first 14 monitoring cycles were used as the training dataset to develop the prediction models, while the last 4 cycles were reserved for evaluating their accuracy. The results show the absolute errors of the ANN are determined to be small, within the range of 10 mm; while it is of 30 mm for EF method. Additionally, the ANN model demonstrated an accuracy improvement of approximately 10% over the EF method. These findings highlight the potential of the ANN model as a reliable and accurate tool for predicting surface subsidence in mining areas.

This study presents a new framework that combines Bayesian inference, Artificial Neural Networks (ANNs), and numerical and experimental modal analysis to determine material properties in structural plates and detect flaws. The Smoothed Finite Element Method (SFEM) is used as the forward solver along with experimental modal analysis to accurately characterize the plate’s material properties. The material properties derived through Bayesian inference are used to create a natural frequency dataset for predicting flaws within the plate. This dataset is used to train ANNs, which are the primary tool for solving the inverse problem. Two different ANN models are developed to predict a complex star-shaped flaw and multiple circular flaws with an accuracy exceeding 96%. This comprehensive framework not only improves flaw detection capabilities but also demonstrates the effectiveness of SFEM in Bayesian inference for material parameter estimation and flaw data generation. The proposed methodology shows significant promise for advancing structural health monitoring and maintenance practices.

Biohydrogen production from potato processing wastewater using double-chamber microbial electrolysis cell with Ru and Pd coated graphite cathodes.

High-entropy alloys (HEAs) are highly concentrated, multicomponent alloys that have received significant attention due to their superior properties compared to conventional alloys. The mechanical properties and hardness are interrelated, and it is widely known that the hardness of HEAs depends on the principal alloying elements and their composition. Therefore, the desired hardness prediction to develop new HEAs is more interesting. However, the relationship of these compositions with the HEA hardness is very complex and nonlinear. In this study, we develop an artificial neural network (ANN) model using experimental data sets (535). The compositional elements—Al, Co, Cr, Cu, Mn, Ni, Fe, W, Mo, and Ti—are considered input parameters, and hardness is considered as an output parameter. The developed model shows excellent correlation coefficients (Adj R2) of 99.84% and 99.3% for training and testing data sets, respectively. We developed a user-friendly graphical interface for the model. The developed model was used to understand the effect of alloying elements on hardness. It was identified that the Al, Cr, and Mn were found to significantly enhance hardness by promoting the formation and stabilization of BCC and B2 phases, which are inherently harder due to limited active slip systems. In contrast, elements such as Co, Cu, Fe, and Ni led to a reduction in hardness, primarily due to their role in stabilizing the ductile FCC phase. The addition of W markedly increased the hardness by inducing severe lattice distortion and promoting the formation of hard intermetallic compounds.

Recurrent spontaneous abortion (RSA) represents a significant challenge in reproductive obstetrics, affecting approximately 5% of couples globally. Despite various treatments, the effectiveness of these interventions remains highly contentious. Emerging evidence suggests that cellular senescence plays a critical role in the pathogenesis of multiple diseases, potentially implicating its involvement in RSA as well. This study integrated two RNA sequencing datasets and employed Weighted Gene Co-Expression Network Analysis (WGCNA) alongside five machine learning algorithms (XGBoost, Boruta, LASSO, SVM-RFE, and Random Forest) to identify cellular senescence genes linked to RSA. Gene expression was validated at both the transcriptome and protein levels using qPCR, Western blot, and immunofluorescence techniques. An Artificial Neural Network (ANN) model was implemented to evaluate their diagnostic value. Additionally, functional enrichment and single-cell RNA sequencing analyses were conducted to investigate the biological functions of these genes. Three cellular senescence genes—TBX2, SRSF3, and TNRC6B—were identified and found to be upregulated in RSA patients. Functional enrichment analysis revealed these genes’ involvement in the MAPK signaling pathway, ECM-receptor interaction, and cell-cell communication. Single-cell RNA sequencing demonstrated the distribution of these genes across various cell types, underscoring their significance in RSA. Furthermore, drug sensitivity analysis identified potential small molecule therapeutics for RSA. Cellular senescence may play a central role in the pathology of RSA, with TBX2, SRSF3, and TNRC6B emerging as potential diagnostic biomarkers. This research enhances our understanding of the molecular mechanisms underlying RSA and lays the groundwork for new diagnostic and therapeutic strategies. Nonetheless, further experimental studies are required to elucidate the specific roles and mechanisms of these genes in RSA, with the ultimate goal of achieving precise prevention and personalized treatment.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-19251-9.

Adhesive wear has been identified as a significant cause of material loss, representing a substantial challenge across diverse industrial sectors. In order to address this issue, it is imperative to conduct studies with the aim of mitigating this degradation. The present study focuses on achieving a high-quality product with minimal mass loss during adhesive wear by utilizing gas nitriding treatment to optimize the wear parameters of AISI 4140 steel. The present study employed the Taguchi methodology and response surface methodology (RSM) in order to design the experiments. A comprehensive investigation was conducted into the key wear parameters, encompassing sliding speed (V), normal load (FN), and the microhardness of nitrided parts (HV). Furthermore, an artificial neural network (ANN) prediction model was developed to forecast the wear performance of 4140 Steel. The ANN model demonstrated an accuracy of approximately 99% when compared to the experimental data. In order to enhance the precision of wear estimation, prediction optimization was conducted using Bayesian and genetic algorithms. The findings demonstrated that the predicted R2 values exhibited a reasonable alignment with the adjusted R2 values, with a discrepancy of less than 0.2. The analysis demonstrated that the normal load is the most significant factor influencing wear, followed by hardness. In contrast, sliding speed was found to have the least significant impact.

The purpose of this study was to investigate the relationship between coagulation related genes (CRGs) and breast cancer (BC). First, we found that most CRGs are abnormally expressed in BC patients and correlated with their prognosis. Therefore, we explored the expression of CRGs in benign and malignant breast tissues in the Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx), extracted differentially expressed CRGs, and established an artificial neural network (ANN) diagnostic model to distinguish the nature of breast tissues, as well as a risk scoring model for prognostic assessment and risk stratification. The specimen transcriptomic data we provided confirmed the diagnostic performance of the ANN model described above. For the risk score model, we used internal and external validation, using ROC curves and C-index values to test its predictive value in the TCGA and Gene Expression Omnibus (GEO) cohorts, and further established a prognostic nomogram for clinical application. In addition, we evaluated the performance of diagnostic and prognostic models using 3 cross-validations methods. RABIF was further identified as a core gene. We performed a more detailed study of RABIF: RT-qPCR of BC cell lines and immunohistochemical staining (IHC) of breast tissue samples showed that RABIF is highly expressed in BC especially in advanced BC. Our study demonstrates the value of CRGs as diagnostic and prognostic targets and may contribute to clinical decision-making in BC.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-19290-2.

In the Petroleum industry, pressure losses in tubing installations must be determined accurately. Traditionally, flowing bottom-hole pressure was determined using mechanical down-hole gauges, this procedure is not cost-effective and less efficient as mechanical tools are prone to damage. This research aims to compare an improved mechanistic model of pressure determination with a machine-learning model that predicted bottom-hole pressure readings. Guo’s mechanistic model was modified in this study while considering some assumptions that affect the estimation. A pressure gradient expression was obtained, and it was solved using a piece-wise iteration approach. The machine learning model was based on an Artificial Neural Network algorithm to predict and further improve the accuracy of the prediction while considering a large production dataset from different wells of the field. In developing the model, the initial dataset was pre-processed to about 2,500 data points; the model was trained, tested, and cross-validated based on the parameters from the data. The results obtained from the mechanistic model gave an accuracy of 0.888 when tested on a fraction of the Volve dataset, while the Artificial Neural Network model gave an accuracy of 0.999 on the test dataset. Finally, this shows that, apart from the ability of machine learning to handle large datasets, it also predicted a high value of accuracy when compared to the improved mechanistic model.

The determination of rate constants for barrierless reactions poses severe problems from a theoretical perspective. The main challenges concern the proper description of the electronic structure of the reacting system, which may have multireference character, the anharmonicity of the relative motions of the fragments, and the proper definition of the reaction coordinate. The literature state of the art in the context of transition state theory is its variable reaction coordinate implementation (VRC-TST), which overcomes these difficulties in determining the number of transition state ro-vibrational states through a Monte Carlo sampling of the potential energy surface (PES) defined by the relative orientation of the two fragments. Although approaching the accuracy of experiments, VRC-TST requires tens of thousands of single-point energy (SPE) evaluations, thus being computationally demanding. The approach developed in this work, named NN-VRCTST, aims at fitting the PES with physics-inspired artificial neural network (ANN) models to be used as surrogate potentials in VRC-TST simulations. The ANN efficacy is evaluated in the computation of high-pressure limit rate constants for gas-phase barrierless reactions and validated over state-of-the-art VRC-TST simulations. It is shown that the NN-VRCTST tool reaches an accuracy within 20% with respect to VRC-TST simulations performed by using traditional approaches. While lowering the number of SPE needed by at least a factor of 4, the computational framework devised here allows one to decouple ANN training and VRC-TST calculations, enabling the optimization of the SPE evaluations as well as the quality inspection of the employed data points. We believe that the NN-VRCTST approach has the potential to evolve into a robust and computationally efficient framework for performing VRC-TST calculations for barrierless reactions.

Abstract Purpose When considering the influence of recognition-based heuristics on entrepreneurs’ strategic decision-making (ESDM), especially in emerging markets, conventional theories and literature on entrepreneurs’ management approach are notably sparse. This study investigates how recognition-based heuristics influence ESDM, particularly in an emerging Asian economy. Design/methodology/approach Data was collected through a survey completed by 237 owners and senior managers of small and medium-sized enterprises (SMEs) in the service, trade, and manufacturing sectors located in the Pakistani cities of Rawalpindi and Islamabad (twin cities). Data was collected using a convenient purposive sampling technique and snowball sampling method. A structural equation modeling-artificial neural network (SEM-ANN) based approach was applied to evaluate the role of recognition-based heuristic predictors. The results were authenticated using regression analysis. Findings The results indicate that recognition-based heuristics—such as alphabetical order, name fluency, and name memorability—have a positive impact on ESDM. This means that recognition-based heuristics are useful tools for entrepreneurs in strategic decision-making. Entrepreneurs who use recognition-based heuristics are more likely to make effective strategic decisions. The ANN results reveal that name memorability has the highest predictive power in positively influencing ESDM, suggesting that memorability plays a crucial role in facilitating more efficient and informed strategic choices. Originality/value This study pioneers research examining the connection between recognition-based heuristics—alphabetical order, name fluency and name memorability—and ESDM in an emerging Asian market. This study contributes to the entrepreneurial management field, particularly regarding the role of recognition-based heuristics in strategic decision-making. This research area is still in its early stages, even in developed economies, and very little work has been conducted in emerging economies. This study makes a significant contribution to the literature in this field. We employed a novel SEM-ANN based evaluation approach that combines the strengths of SEM and ANN. This integration allows for a comprehensive analysis of both linear and nonlinear relationships between variables, providing a nuanced understanding of the complex dynamics involved in ESDM, and differentiating this study from other studies in the field.

The detection of potential biological threat elements is of vital importance in terms of environmental monitoring, public health, and public safety. The possibility of future military or terrorist use of such biological agents poses a serious risk to global security. Therefore, early detection of the threat plays a critical role in taking effective measures against a possible biological attack and putting emergency action plans into effect on time. In this study, an electronic nose system that can safely and effectively identify complex gas mixtures was designed by developing an artificial intelligence-based model on the data collected using a sensitive gas sensor matrix. However, in research studies, the direct use of highly hazardous biological and chemical agents is not possible due to high-security risks, ethical concerns, and legal restrictions. Therefore, in this study, a simulation environment was established to represent complex biological and chemical gas elements. The collected data was analysed with the Artificial Neural Network model, which is known to show strong performance in gas recognition tasks. The findings indicate that this approach can be used to detect potential biological threats and that electronic nose technologies can be evaluated in the field of security with artificial intelligence-supported applications.

Abstract Due to the increasing global demand for sustainable energy and the variable nature of solar radiation, accurate forecasting of photovoltaic (PV) system performance has become essential. This study focuses on the development of an artificial neural network (ANN) model to predict monthly electricity production in a rooftop photovoltaic power plant. The model uses meteorological inputs such as direct normal irradiation, diffuse radiation, and average monthly temperature, covering the period from January 2017 to December 2023. The ANN demonstrated high predictive accuracy and reliability, making it a valuable tool for energy management, planning, and integration of PV systems into power grids. While the dataset spans 2017–2023, the model is structured to allow generalization for future forecasting applications under similar input conditions.

Dissolved oxygen (DO) is a critical indicator of estuarine ecosystem health, influenced by complex physicochemical interactions. This study employs Artificial Neural Networks (ANNs) and Adaptive Neuro-Fuzzy Inference System (ANFIS) to predict DO concentrations in the Bouregreg Estuary, Morocco, using salinity, temperature, pH, and transparency data collected on 10/11/2021, 20/03/2022, and 02/07/2022 at three stations during high and low tides. The dataset includes DO concentrations (5.5–11.8 mg/L), salinity (19.4–35.8 ppt), temperature (17–23 °C), pH (7.0–8.7), and transparency (16–21 cm). Three ANN models (Back Propagation Neural Network [BPNN], Generalized Regression Neural Network [GRNN], Recurrent Neural Network [RNN]), and four ANFIS models (with triangular, trapezoidal, Gaussian, and Generalized Bell membership functions) were developed. The ANFIS model with Gaussian membership function (ANFIS4) achieved the highest performance in Scenario II (R 2 = 0.9500, RMSE = 0.2800 mg/L, and PBIAS = −1.2000%). Temperature and salinity were key predictors, with salinity capturing estuarine mixing dynamics. Comprehensive visualizations of model validation, spatial-temporal variations, and parameter relationships enhance the understanding of DO dynamics. These findings provide a robust framework for estuarine management and highlight the role of tidal mixing in oxygen distribution.

Additive manufacturing has promising potential for the development of 3D-printed protective structures such as stab-resistant body armor. However, no research to date has examined the impact of 3D printing parameters on the protective performance of such 3D-printed structures manufactured using fused filament fabrication technology. This study, therefore, investigates the effects of five key printing parameters: layer thickness, print speed, print temperature, infill density (Id), and layer width, on the mechanical and protective performance of 3D-printed polycarbonate (PC) armor. A Taguchi L27 matrix was employed to systematically analyze these parameters, with toughness, stab penetration depth, and armor panel weight as the primary responses. ANOVA results, along with the Taguchi approach, demonstrated that Id was the most influential factor across all print parameters. This is because a higher Id led to denser structures, reduced voids and porosities, and enhanced energy absorption, significantly increasing toughness while reducing penetration depth. Morphological analysis supported the statistical findings regarding the role of Id on the performance of such structures. With optimized printing parameters, no penetration to the armor panels was recorded, outperforming the UK body armor standard of a maximum permitted knife penetration depth of 8 mm. Moreover, an artificial neural network (ANN) utilizing the 5-14-12-3 topology was created to predict the toughness, stab penetration depth, and armor panel weight of 3D-printed armors. The ANN model demonstrated better prediction performance for stab penetration depth compared to the Taguchi method, confirming the successful application of such an approach. These findings provide a critical foundation for the development of high-performance 3D-printed protective structures.

Naltrexone Induced Agitation Management: Employing a Hybrid Artificial Neural Network Model to Determine the Appropriate Dosage of Intravenous Diazepam

This research optimizes an Artificial Neural Network (ANN) model using Grid Search (GS) for predicting the rice yields in Indonesia. The purpose of this research was to enhance the performance of the ANN model by systematically tuning its hyperparameters to improve its predictive accuracy. This research uses the Multilayer Perceptron (MLP) method, and a comprehensive GS method was employed to explore various hyperparameter combinations, including the number of hidden layers, activation functions, solvers, regularization parameters, and learning rates. The optimization process involved evaluating each hyperparameter configuration using cross-validation to select the best model based on performance metrics, including the coefficient of determination (R²), Mean Absolute Error (MAE), and Mean Squared Error (MSE). The study's results indicate that the optimized ANN model achieved an R² of 97.41%, MAE of 766.69, and MSE of 1859857.06, outperforming the model without hyperparameters. This study highlights the effectiveness of the GS optimization in enhancing the ANN model performance, demonstrating that Hyperparameter Tuning (HT) is crucial for achieving improved prediction accuracy. This study concludes that the ANN model can be optimized for practical use in predicting the rice yields, as it shows strong performance.

Accurate forecasting of prices of agricultural commodities has paramount importance, as it enables farmers, policymakers, and the government to make well-informed decisions. While stochastic models like autoregressive integrated moving average (ARIMA) and its components have gained popularity in modelling linear dynamics, they fall short when it comes to capturing the inherent nonlinearity present in the datasets. Machine learning (ML) techniques, such as artificial neural networks (ANNs), have rapidly gained prominence in the field of forecasting, as they are better suited to handle the nonlinearity present in the data. Therefore, a crucial step considered was the preprocessing of the time series data to extract the underlying signal. The present studies focus on the application of the hybrid models and also analyses the prediction accuracy of different models in forecasting monthly prices of maize in Andhra Pradesh, India. The models used for the study were ARIMA, GARCH, ANN, waveletARIMA, waveletGARCH and waveletANN models were employed to compare the accuracy performance of different models in price forecasting. Empirical evidence clearly demonstrates that substantial improvements over conventional techniques were achieved by implementing a wavelet-based combination approach in conjunction with machine learning (ML) techniques. This approach capitalises on the strengths of both wavelet transformations and ML algorithms, resulting in enhanced forecasting performance and accuracy. The ARIMA (2, 1, 1) model was identified as one of the appropriate models to forecast the prices of maize in Andhra Pradesh. The significant p- value indicates the strength of evidence against the null hypothesis. A small p-value in ARCH-LM test reveals the presence of conditional heteroskedasticity. The non-linear time series, NNAR (3-10-1) model was identified as the best model, 51 weights with linear output units. The study revealed that the wavelet based hybrid models forecasted the maize price series better than the individual stochastic models, confirmed based on the performance metrics. Among the proposed models the best performed model identified was waveletANN, with the least MSE and RMSE values.

Ongoing time overruns in urban Flexible Pavement Projects (FPP) highlight the inadequacy of traditional risk forecasting techniques, which often overlook nonlinear and project-specific delay factors. While recent Artificial Intelligence (AI)-based approaches have been proposed, most remain at a descriptive level, demonstrating only a few mathematically expressible and experimentally validated models suitable for urban road networks. This study addresses these gaps by developing a closed-form Artificial Neural Network (ANN) model using nine carefully selected predictors drawn from recent engineering practices and project data in Najaf, Iraq. The model incorporates advanced preprocessing, including robust outlier detection and min–max scaling, and is trained on a newly compiled dataset covering 35 major projects, thereby improving on previous studies' shortcomings in terms of both data quality and methodological transparency. Empirical results demonstrate that the ANN substantially outperforms baseline models, achieving an R2 of 0.847 and a Mean Absolute Percentage Error (MAPE) of 7.10%, with all improvements being statistically significant (p < 0.001). Additionally, feature sensitivity analysis identified payment delay and contractor experience as the most influential risk factors, underscoring the model's practical relevance. Importantly, the modular mathematical structure of the ANN facilitates transparent benchmarking and direct transferability to other urban regions, while creating a sound and replicable paradigm for impact-based, data-driven decision-making and planning infrastructure. Thus, the proposed model constitutes a benchmark for future research on predictive modelling of time overruns in urban pavement projects.

Integrating disease severity with real-time meteorological variables and advanced machine learning techniques has provided valuable predictive insights for assessing disease severity in wheat. This study emphasizes the potential of machine learning models, particularly artificial neural networks (ANN), in predicting wheat disease severity with high accuracy. The field experiment was conducted over two consecutive rabi growing seasons (2023 And 2024) using a randomized block design with four sowing dates to investigate critical weather-disease relationships for two key wheat pathogens: Puccinia striiformis f. sp. tritici (yellow rust) and Blumeria graminis f. sp. tritici (powdery mildew). Weekly assessments of disease severity were combined with meteorological data and analyzed using ANN and regularized regression models. The ANN model demonstrated superior predictive accuracy for yellow rust and powdery mildew, achieving R-squared values (R2 of 0.96 And 0.98 for calibration And 0.93 And 0.95 for validation, respectively. Random Forest (RF) models also exhibited robust performance with R2 values of 0.97 And 0.98 for calibration And 0.93 And 0.90 for validation for yellow rust and powdery mildew, respectively. In contrast, Elastic Net, Lasso, and Ridge regression models showed comparatively moderate predictive capabilities. Principal component analysis (PCA) explained the key meteorological variables influencing disease incidence, with evapotranspiration, temperature, wind speed and humidity emerging as critical factors. Disease prediction is an important aspect of developing a decision support system, and it makes farmers make informed decisions to optimize production.