Predictive Capability Of Model Research Articles

Wastewater quality prediction based on channel attention and TCN-BiGRU model.

Water quality prediction is crucial for water resource management, as accurate forecasting can help identify potential issues in advance and provide a scientific basis for sustainable management. To predict key water quality indicators, including chemical oxygen demand (COD), suspended solids (SS), total phosphorus (TP), pH, total nitrogen (TN), and ammonia nitrogen (NH₃-N), we propose a novel model, CA-TCN-BiGRU, which combines channel attention mechanisms with temporal convolutional networks (TCN) and bidirectional gated recurrent units (BiGRU). The model, which uses a multi-input multi-output (MIMO) architecture, is capable of simultaneously predicting multiple water quality indicators. It is trained and tested using data from a wastewater treatment plant in Huizhou, China. This study investigates the impact of data preprocessing and the channel attention mechanism on model performance and compares the predictive capabilities of various deep learning models. The results demonstrate that data preprocessing significantly improves prediction accuracy, while the channel attention mechanism enhances the model's focus on key features. The CA-TCN-BiGRU model outperforms others in predicting multiple water quality indicators, particularly for COD, TP, and SS, where MAE and RMSE decrease by approximately 23% and 26%, respectively, and R2 improves by 5.85%. Moreover, the model shows strong robustness and real-time performance across different wastewater treatment plants, making it suitable for short-term (1-3days) water quality prediction. The study concludes that the CA-TCN-BiGRU model not only achieves high accuracy but also offers low computational overhead and fast inference speed, making it an ideal solution for real-time water quality monitoring.

Mfgnn: Multi-Scale Feature-Attentive Graph Neural Networks for Molecular Property Prediction.

In the realm of artificial intelligence-driven drug discovery (AIDD), accurately predicting the influence of molecular structures on their properties is a critical research focus. While deep learning models based on graph neural networks (GNNs) have made significant advancements in this area, prior studies have primarily concentrated on molecule-level representations, often neglecting the impact of functional group structures and the potential relationships between fragments on molecular property predictions. To address this gap, we introduce the multi-scale feature attention graph neural network (MfGNN), which enhances traditional atom-based molecular graph representations by incorporating fragment-level representations derived from chemically synthesizable BRICS fragments. MfGNN not only effectively captures both the structural information of molecules and the features of functional groups but also pays special attention to the potential relationships between fragments, exploring how they collectively influence molecular properties. This model integrates two core mechanisms: a graph attention mechanism that captures embeddings of molecules and functional groups, and a feature extraction module that systematically processes BRICS fragment-level features to uncover relationships among the fragments. Our comprehensive experiments demonstrate that MfGNN outperforms leading machine learning and deep learning models, achieving state-of-the-art performance in 8 out of 11 learning tasks across various domains, including physical chemistry, biophysics, physiology, and toxicology. Furthermore, ablation studies reveal that the integration of multi-scale feature information and the feature extraction module enhances the richness of molecular features, thereby improving the model's predictive capabilities.

A review of integrated use of machine learning algorithms, GIS and remote sensing techniques in the prediction of rainfall patterns and floods in the U. S.

This paper systematically reviews the integrated use of machine learning algorithms, Geographic Information Systems (GIS), and Remote Sensing (RS) techniques in the prediction of rainfall patterns and flood events in the U.S. With increasing climate variability, accurate forecasting of rainfall and flood risks has become critical to safeguarding communities and infrastructure. GIS enables spatial analysis and mapping of flood-prone areas, supporting risk assessment and disaster preparedness. RS contributes real-time satellite imagery and environmental data, essential for tracking rainfall patterns and assessing surface conditions. Machine learning algorithms enhance these technologies by providing predictive modeling capabilities, allowing for more precise forecasts of rainfall intensity and flood potential. This paper explores the synergy between GIS, RS, and machine learning, emphasizing their combined impact on improving flood prediction accuracy and decision-making in disaster management. Key challenges, including data heterogeneity, computational demands, and the integration of diverse datasets, are discussed. Additionally, the paper reviews current U.S. policies on data-sharing and technology adoption, highlighting the need for regulatory frameworks that support innovation while ensuring data privacy and accuracy. Through an analysis of recent studies, this paper presents a comprehensive overview of the advantages and constraints of using these integrated technologies for flood prediction, offering insights into future directions and recommendations for enhancing flood management systems. The review concludes that advancing integrated GIS, RS, and machine learning applications will require addressing data-related challenges and fostering collaborative efforts across agencies to strengthen flood prediction and resilience capabilities in the U.S.

Critical Appraisal of Integrated CFD/Surface Roughness Models for Additive Manufactured Swirl Burners

Abstract Additive manufacturing technology creates complex parts that are impossible to be manufactured by traditional methods. However, the process often results in rough and irregular surfaces that can affect performance. Here Computational Fluid Dynamics (CFD) is considered to optimise component design for use in applications such as gas turbine. However, modelling the interactions between turbulent flows and Additive Manufacture (AM)-generated wall roughness affects the predictive capability of numerical models due to difficulty in thoroughly characterising rough wall texture. To address this issue, this study aims to appraise two common wall roughness approaches within the RANS framework: the modified law-of-the-wall and roughness-resolving approaches. The modified law-of-the wall is based on the correlation that converts the measured surface roughness parameters to the equivalent sand-grain roughness height. The latter approach involves the resolution of the roughness elements within the computational grid. The simulations were validated and compared against the velocity data published for the burner with AM swirl nozzle inserts of different surface finishes. The Realizable k-ɛ turbulence model was selected for all the CFD simulations, as it has been already validated for swirling flows. The results showed that the roughness-resolving approach was more accurate than the modified law-of-the wall correlation, with a good match with the experimental velocity data, predicting the velocity shift to the center. The model also revealed the shortened recirculation zone with increasing surface roughness. This is important in predicting flame stability and emissions performance.

A preoperative predictive model based on multi-modal features to predict pathological complete response after neoadjuvant chemoimmunotherapy in esophageal cancer patients

BackgroundThis study aimed to develop a multi-modality model by incorporating pretreatment computed tomography (CT) radiomics and pathomics features along with clinical variables to predict pathologic complete response (pCR) to neoadjuvant chemoimmunotherapy in patients with locally advanced esophageal cancer (EC).MethodA total of 223 EC patients who underwent neoadjuvant chemoimmunotherapy followed by surgical intervention between August 2021 and December 2023 were included in this study. Radiomics features were extracted from contrast-enhanced CT images using PyrRadiomics, while pathomics features were derived from whole-slide images (WSIs) of pathological specimens using a fine-tuned deep learning model (ResNet-50). After feature selection, three single-modality prediction models and a combined multi-modality model integrating two radiomics features, 11 pathomics features, and two clinicopathological features were constructed using the support vector machine (SVM) algorithm. The performance of the models were evaluated using receiver operating characteristic (ROC) analysis, calibration plots, and decision curve analysis (DCA). Shapley values were also utilized to explain the prediction model.ResultsThe predictive capability of the multi-modality model in predicting pCR yielded an area under the curve (AUC) of 0.89 (95% confidence interval [CI], 0.75-1.00), outperforming the radiomics model (AUC 0.70 [95% CI 0.54-0.85]), pathomics model (AUC 0.77 [95% CI 0.53-1.00]), and clinical model (AUC 0.63 [95% CI 0.46-0.80]). Additionally, both the calibration plot and DCA curves support the clinical utility of the integrated multi-modality model.ConclusionsThe combined multi-modality model we propose can better predict the pCR status of esophageal cancer and help inform clinical treatment decisions.

Investigating the Relationship between Soil Gas Radon and Soil Permeability by Using Artificial Neural Networks

This study aims to explore the relationship between soil gas radon concentration (CRn) and soil permeability (k). To accomplish this, a single linear regression analysis (SLRA) model and an artificial neural network (ANN) model were built from 142 soil gas CRn and k measurements collected from the literature. When soil gas CRn values predicted by both models were compared with those measured, the ANN model outperformed the SLRA model. Furthermore, several performance metrics, including correlation coefficient, root mean square error, relative absolute error, and mean absolute error were determined to examine the prediction capabilities of SLRA and ANN models. The metrics obtained demonstrated that the ANN model exhibited superior performance to the SLRA model, thereby showing the accuracy and applicability of the ANN model for forecasting soil gas CRn values. The study's findings indicated that the developed ANN model may be utilized to forecast soil gas CRn values based on soil k values.

A Review of Ignition Characteristics and Prediction Model of Combustor Under High-Altitude Conditions

High-altitude relight is a critical challenge for aero-engines, directly impacting the safety and emergency response capabilities of aircraft. This paper systematically reviews the physical mechanisms, key factors, and relevant prediction models of high-altitude relight, highlighting the detrimental effects of extreme conditions such as low pressure and temperature on fuel evaporation rates, flame propagation speeds, and turbulent combustion processes. A comprehensive overview of the current state of high-altitude relight research is presented, alongside recommendations for enhancing the ignition performance of aero-engines under extreme conditions. This paper focuses on the development of ignition prediction models, including early empirical and semi-empirical models, as well as physics-based models for turbulent flame propagation and flame kernel tracking, assessing their applicability in high-altitude relight scenarios. Although flame kernel tracking has shown satisfactory performance in predicting ignition probability, it still overly relies on manually set parameters and lacks precise descriptions of the physical processes of flame kernel generation. Future studies on some topics, including refining flame kernel modeling, strengthening the integration of experimental data and numerical simulations, and exploring the incorporation of new ignition technologies, are needed, to further improve model reliability and predictive capability.

Characterization of uncertainties in electron-argon collision cross sections

Abstract The predictive capability of a plasma discharge model depends on accurate representations of electron-impact collision cross sections, which determine the corresponding reaction rates and electron transport properties. The values of cross sections can be known only approximately either through experiments or simulations and are thus subject to uncertainties. Quantifying the uncertainties in plasma simulations allows us to assess the reliability of simulations and to provide a basis for interpreting discrepancies between simulations and experiments. For such uncertainty quantification of plasma simulations, it is essential to quantify the uncertainties of the underlying cross sections. Although much effort has been committed to calibrate the cross section values, their uncertainties are not well investigated. We characterize uncertainties in electron-argon atom collision cross sections using a Bayesian framework. Six collision processes—elastic momentum transfer, ionization, and four excitations—are characterized with semi-empirical models, which effectively capture the features important to the macroscopic properties of the plasma. A probability model for the uncertain parameters of these semi-empirical models is developed. Specifically, a Gaussian-process likelihood model is proposed to capture discrepancies among data sets, as well as the model-form inadequacies of the semiempirical models. Two other likelihood models are compared with the proposed Gaussian-process model, to illustrate the importance of the choice of the likelihood model. The cross section models are calibrated using the electron-beam experiments and ab-inito quantum simulations. The resulting calibrated uncertainties capture well the scattering among the data sets. The calibrated cross section models are further validated against swarm-parameter experiments and zero-dimensional Boltzmann equation simulations of widely used cross section datasets.

Comparative Analysis of Recurrent Neural Networks with Conjoint Fingerprints for Skin Corrosion Prediction.

Skin corrosion assessment is an essential toxicity end point that addresses safety concerns for topical dosage forms and cosmetic products. Previously, skin corrosion assessments required animal testing; however, differences in skin architecture and ethical concerns regarding animal models have fostered the advancement of alternative methods such as in silico and in vitro models. This study aimed to develop deep learning (DL) models based on recurrent neural networks (RNNs) for classifying skin corrosion of chemical compounds based on chemical language notation, molecular substructure, physicochemical properties, and a combination of these three properties called conjoint fingerprints. Simple RNN, long short-term memory, bidirectional long short-term memory (BiLSTM), gated recurrent units, and bidirectional gated recurrent units models, along with 11 molecular features, were employed to generate 55 RNN-based models. Applicability domain and permutation importance analysis were exploited for additional trustable prediction and explanation ability of the models, respectively. Our findings indicate that BiLSTM with conjoint features of MACCS keys and physicochemical descriptors is the most effective model with 84.3% accuracy, 89.8% area under the curve, and 57.6% Matthews correlation coefficient for the external test performance. Furthermore, our model accurately predicted the skin corrosion toxicity of all new and unseen compounds beyond our test set, highlighting prominent classification performance compared to existing skin corrosion models. This finding will contribute to the utilization of DL and conjoint characteristics of molecular structure to enhance the model's predictive capability for skin toxicity assessment.

Digital Image Correlation and Numerical Analysis of Mechanical Behavior in Photopolymer Resin Lattice Structures.

Cellular structures are increasingly utilized in modern engineering due to their exceptional mechanical and physical properties. In this study, the deformation and failure mechanisms of two energy-efficient lattice structures-hexagonal honeycomb and re-entrant honeycomb-were investigated. These structures were manufactured using additive stereolithography with light-curable Durable Resin V2. The experimental testing of the topologies under two perpendicular loading directions employed the 3D Digital Image Correlation (DIC) system to capture strain fields and deformation patterns, providing insights into structural behavior and failure mechanisms. The unit cells of the topologies were scaled up to enable precise optical measurements while preserving their structural interaction characteristics. Numerical simulations, conducted using the SAMP-1 material model in LS-DYNA and calibrated with tensile and compression test data, accurately replicated the behavior of the studied topologies and demonstrated good agreement with experimental results. The hexagonal structure, loaded along axis 2, showed the best fit, with deviations within 5%, while the re-entrant honeycomb structure exhibited weaker yet reasonable agreement. By integrating experimental and numerical approaches, the research validates the SAMP-1 model's predictive capabilities for lattice structures and provides a framework for analyzing energy-absorbing lattice topologies.

The landscape of plasma proteomic links to human organ imaging.

Plasma protein levels provide important insights into human disease, yet a comprehensive assessment of plasma proteomics across organs is lacking. Using large-scale multimodal data from the UK Biobank, we integrated plasma proteomics with organ imaging to map their phenotypic and genetic links, analyzing 2,923 proteins and 1,051 imaging traits across multiple organs. We uncovered 5,067 phenotypic protein-imaging associations, identifying both organ-specific and organ-shared proteomic relations, along with their enriched protein-protein interaction networks and biological pathways. By integrating external gene expression data, we observed that plasma proteins associated with the brain, liver, lung, pancreas, and spleen tended to be primarily produced in the corresponding organs, while proteins associated with the heart, body fat, and skeletal muscle were predominantly expressed in the liver. We also mapped key protein predictors of organ structures and showed the effective stratification capability of plasma protein-based prediction models. Furthermore, we identified 8,116 genetic-root putative causal links between proteins and imaging traits across multiple organs. Our study presents the most comprehensive pan-organ imaging proteomics map, bridging molecular and structural biology and offering a valuable resource to contextualize the complex roles of molecular pathways underlying plasma proteomics in organ systems.

High-risk habitat radiomics model based on ultrasound images for predicting lateral neck lymph node metastasis in differentiated thyroid cancer

BackgroundThis study aims to evaluate the predictive usefulness of a habitat radiomics model based on ultrasound images for anticipating lateral neck lymph node metastasis (LLNM) in differentiated thyroid cancer (DTC), and for pinpointing high-risk habitat regions and significant radiomics traits.MethodsA group of 214 patients diagnosed with differentiated thyroid carcinoma (DTC) between August 2021 and August 2023 were included, consisting of 107 patients with confirmed postoperative lateral lymph node metastasis (LLNM) and 107 patients without metastasis or lateral cervical lymph node involvement. An additional cohort of 43 patients was recruited to serve as an independent external testing group for this study. Patients were randomly divided into training and internal testing group at an 8:2 ratio. Region of interest (ROI) was manually outlined, and habitat analysis subregions were defined using the K-means method. The ideal number of subregions (n = 5) was determined using the Calinski-Harabasz score, leading to the creation of a habitat radiomics model with 5 subregions and the identification of the high-risk habitat model. Area under the curve (AUC) values were calculated for all models to assess their validity, and predictive model nomograms were created by integrating clinical features. The internal and external testing dataset is employed to assess the predictive performance and stability of the model.ResultsIn internal testing group, Habitat 3 was identified as the high-risk habitat model in the study, showing the best diagnostic efficacy among all models (AUC(CRM) vs. AUC(Habitat 3) vs. AUC(CRM + Habitat 3) = 0.84(95%CI:0.71–0.97) vs. 0.90(95%CI:0.80-1.00) vs. 0.79(95%CI:0.65–0.93)). Moreover, integrating the Habitat 3 model with clinical features and constructing nomograms enhanced the predictive capability of the combined model (AUC = 0.95(95%CI:0.88-1.00)). In this study, an independent external testing cohort was utilized to assess the model’s accuracy, yielding an AUC of 0.88 (95%CI: 0.78–0.98).ConclusionThe integration of the High-Risk Habitats (Habitat 3) radiomics model with clinical characteristics demonstrated a high predictive accuracy in identifying LLNM. This model has the potential to offer valuable guidance to surgeons in deciding the necessity of LLNM dissection for DTC.Clinical trial numberNot applicable.

Unraveling the potential mechanism and prognostic value of pentose phosphate pathway in hepatocellular carcinoma: a comprehensive analysis integrating bulk transcriptomics and single-cell sequencing data.

Hepatocellular carcinoma (HCC) remains a malignant and life-threatening tumor with an extremely poor prognosis, posing a significant global health challenge. Despite the continuous emergence of novel therapeutic agents, patients exhibit substantial heterogeneity in their responses to anti-tumor drugs and overall prognosis. The pentose phosphate pathway (PPP) is highly activated in various tumor cells and plays a pivotal role in tumor metabolic reprogramming. This study aimed to construct a model based on PPP-related Genes for risk assessment and prognosis prediction in HCC patients. We integrated RNA-seq and microarray data from TCGA, GEO, and ICGC databases, along with single-cell RNA sequencing (scRNA-seq) data obtained from HCC patients via GEO. Based on the "Seurat" R package, we identified distinct gene clusters related to the PPP within the scRNA-seq data. Using a penalized Cox regression model with least absolute shrinkage and selection operator (LASSO) penalties, we constructed a risk prognosis model. The validity of our risk prognosis model was further confirmed in external cohorts. Additionally, we developed a nomogram capable of accurately predicting overall survival in HCC patients. Furthermore, we explored the predictive potential of our risk model within the immune microenvironment and assessed its relevance to biological function, particularly in the context of immunotherapy. Subsequently, we performed in vitro functional validation of the key genes (ATAD2 and SPP1) in our model. A ten-gene signature associated with the PPP was formulated to enhance the prediction of HCC prognosis and anti-tumor treatment response. Following this, the ROC curve, nomogram, and calibration curve outcomes corroborated the model's robust clinical predictive capability. Functional enrichment analysis unveiled the engagement of the immune system and notable variances in the immune infiltration landscape across the high and low-risk groups. Additionally, tumor mutation frequencies were observed to be elevated in the high-risk group. Based on our analyses, the IC50 values of most identified anticancer agents demonstrated a correlation with the RiskScore. Additionally, the high-risk and low-risk groups exhibited differential sensitivity to various drugs. Cytological experiments revealed that silencing ATAD2 or SPP1 suppresses malignant phenotypes, including viability and migration, in liver cancer cells. In this study, a novel gene signature related to the PPP was developed, demonstrating favorable predictive performance. This signature holds significant guiding value for assessing the prognosis of HCC patients and directing individualized treatment strategies.

Predicting stacking fault energy in austenitic stainless steels via physical metallurgy-based machine learning approaches

Stacking fault energy (SFE) significantly influences plastic deformation, strength, and processing performance, making accurate assessment and prediction of SFE essential for material design and optimization. Traditional SFE calculations mainly rely on experimental measurements and thermodynamic theories, with the former usually being time-consuming and the latter limited in applicability at different compositions. To overcome these limitations, this study proposes a machine learning (ML) strategy introducing physical metallurgy (PM) parameters relevant to SFE, aiming to achieve robust predictions. Specifically, this study evaluates three methods for introducing PM information into ML (as an input, an intermediate parameter, and a transfer source), with transfer learning as the best strategy. Initially, various PM parameters were calculated based on alloy composition and temperature, and subsequently used as inputs to train a convolutional neural network (CNN). This source model was then transferred to the SFE prediction model. The results from the model transfer using different PM information show that incorporating phase-transformation driving force (DF) as a source model for SFE prediction provided the most accurate and reliable results. This approach of introducing PM parameters into ML significantly improves the predictive capability of SFE models, offering a new perspective and solution for the prediction of SFE. Furthermore, this method may also be applicable to the prediction of other material properties during material design and optimization.

Fractional and stochastic modeling of breast cancer progression with real data validation.

This study presents a novel approach to modeling breast cancer dynamics, one of the most significant health threats to women worldwide. Utilizing a piecewise mathematical framework, we incorporate both deterministic and stochastic elements of cancer progression. The model is divided into three distinct phases: (1) initial growth, characterized by a constant-order Caputo proportional operator (CPC), (2) intermediate growth, modeled by a variable-order CPC, and (3) advanced stages, capturing stochastic fluctuations in cancer cell populations using a stochastic operator. Theoretical analysis, employing fixed-point theory for the fractional-order phases and Ito calculus for the stochastic phase, establishes the existence and uniqueness of solutions. A robust numerical scheme, combining the nonstandard finite difference method for fractional models and the Euler-Maruyama method for the stochastic system, enables simulations of breast cancer progression under various scenarios. Critically, the model is validated against real breast cancer data from Saudi Arabia spanning 2004-2016. Numerical simulations accurately capture observed trends, demonstrating the model's predictive capabilities. Further, we investigate the impact of chemotherapy and its associated cardiotoxicity, illustrating different treatment response scenarios through graphical representations. This piecewise fractional-stochastic model offers a powerful tool for understanding and predicting breast cancer dynamics, potentially informing more effective treatment strategies.

Investigating the structure-property correlations of pyrolyzed phenolic resin as a function of degree of carbonization.

Carbon-carbon (C/C) composites are attractive materials for high-speed flights and terrestrial atmospheric reentry applications due to their insulating thermal properties, thermal resistance, and high strength-to-weight ratio. It is important to understand the evolving structure-property correlations in these materials during pyrolysis, but the extreme laboratory conditions required to produce C/C composites make it difficult to quantify the properties in situ. This work presents an atomistic modeling methodology to pyrolyze a crosslinked phenolic resin network and track the evolving thermomechanical properties of the skeletal matrix during simulated pyrolysis. First, the crosslinked resin is pyrolyzed and the resulting char yield and mass density are verified to match experimental values, establishing the model's powerful predictive capabilities. Young's modulus, yield stress, Poisson's ratio, and thermal conductivity are calculated for the polymerized structure, intermediate pyrolyzed structures, and fully pyrolyzed structure to reveal structure-property correlations, and the evolution of properties are linked to observed structural features. It is determined that reduction in fractional free volume and densification of the resin during pyrolysis contribute significantly to the increase in thermomechanical properties of the skeletal phenolic matrix. A complex interplay of the formation of six-membered carbon rings at the expense of five and seven-membered carbon rings is revealed to affect thermal conductivity. Increased anisotropy was observed in the latter stages of pyrolysis due to the development of aligned aromatic structures. Experimentally validated predictive atomistic models are a key first step to multiscale process modeling of C/C composites to optimize next-generation materials.

Enhancing hydraulic fracturing efficiency through machine learning

Hydraulic fracturing (HF) is a technique employed in the oil and gas industry to extract hydrocarbon resources from shale formations and low-permeability rocks. This process involves the creation of new fractures and the extension of existing ones within the rock by injecting a high-pressure fracturing fluid, which enhances the flow of hydrocarbons. To assess the efficiency and safety of HF, various factors such as fracture orientation, geometry, length, distribution, and stimulated reservoir volume are analyzed. In the complex field of HF, the accurate assessment of fracture parameters is crucial for optimizing operational efficiency and ensuring environmental safety. This study investigates the role of machine learning (ML) methodologies in enhancing the precision of these assessments, with a particular focus on fracture width a critical factor in the effectiveness of HF. By utilizing a comprehensive dataset from hydrocarbon fields, in this study, the application of several advanced ML models was investigated, including Artificial Neural Networks (ANNs), Random Forest (RF), and K-Nearest Neighbors (KNNs). These models were specifically employed to predict and assess fracture width based on a variety of geological and operational parameters: geometric factors (γ), Shear Modulus (G), Poisson’s ratio (ν), viscosity of the fracturing fluid (µ in centipoise), crack height (h), fluid efficiency (η), injection time (t), and crack length (x). This study evaluated the performance of ANN, RF, and KNN models, achieving accuracies of 0.978, 0.979, and 0.893, respectively, which underscores their strong predictive modeling capabilities. The results were meticulously documented for each methodology. The RMSE values for ANN, KNN, and RF were 7.552, 15.711, and 6.194, respectively. Notably, the RF approach demonstrated superior performance with an RMSE of 6.194, establishing it as the most accurate method. This study highlights the transformative potential of ML in the pre-stimulation planning phase of HF, enabling enhanced real-time decision-making and operational optimization, which ultimately results in more effective and efficient fracturing operations.

Simulation of Rice Performance under Alkaline Soil Pedon Using CERES-Rice Model

A multi-location field experiment was conducted on crop establishment methods of rice under different soil texture. Field experiments in Indian Punjab were conducted on three soil textures including sandy loam, sandy clay loam, and clay loam soils. The experiment was conducted in randomized block design with nine treatment combinations of direct-seeded rice (DSR) and transplanted flooded rice (TFR), along with different irrigation strategies. The results of multi locations field study were used for CERES-Rice model calibration and validation. Simulation study was performed in a soil pedon near Ferozpur, Punjab to evaluate the performance of rice establishment methods under alkaline soil enlivenment. The CERES-Rice model showed satisfactory accuracy in simulating grain yield, biomass, and evapotranspiration (ET), with low RMSE values, indicating minimal residual variation. Other evaluation indices such as Nash-Sutcliffe Modelling Efficiency (ME), R² values demonstrated strong correlation between observed and simulated data. The index of agreement (d) values, ranging from 0.71 to 0.76, indicated good reliability of the model. These results confirmed the model's predictive capability and effectiveness under varying climatic conditions, supporting improved crop management. The simulated grain yield of TFR in neutral soil (Pedon 1) was 5.7 t ha⁻¹ and reduced to 5.0 t ha⁻¹ for direct seeded rice. In contrast, alkaline soil (Pedon 2) had lower yields (4.9 t ha⁻¹ for TFR) and crop failure for DSR, reflecting poor nutrient dynamics and water retention. The water use efficiency for TFR in alkaline soil was slightly reduced, highlighting challenges in such soils.

MACHINE LEARNING-BASED FORECASTING OF BIOACCUMULATION AND HISTOPATHOLOGICAL EFFECTS IN AQUATIC ORGANISMS

Heavy metal contamination in freshwater environments poses significant risks to aquatic organisms and human health, as these heavy metals enter freshwater systems through various sources, including industrial waste, agricultural runoff, mining and atmospheric deposition. Efforts to develop efficient methods for removing heavy metals from wastewater have gained momentum in recent years. This study focuses on machine learning (ML) models for predicting the bioaccumulation and histopathological effects of heavy metal pollutants on aquatic life under various climate change scenarios. The ML models have shown promise in forecasting the impacts of heavy metal pollution on freshwater ecosystems and informing conservation strategies. It is crucial to understand the complex interactions between environmental factors, climate change and ecosystem health. This study discusses the importance of incorporating diverse species and environmental factors in these models and acknowledges potential challenges, such as inaccuracies and data misinterpretation. Enhancing the predictive capabilities of ML models is essential for better environmental management and conservation practices via refinement and validation of models using updated data and advanced methodologies. This study also emphasizes the broad potential of ML in environmental research, improvement of model capabilities and challenges posed by heavy metal pollution and climate change.

Artificial Neural Network Approach for Main Engine Power Prediction of General Cargo Vessels

Before the physical constructing of a ship will start, it must first go through a multistage design process. Determining the ship's main engine power is a critical stage in the concept design process. This work established a model to predict for the main engine power of general cargo ships. The model input parameters included ship length overall, breadth, gross tonnage, DWT and ship service speed. In the training stage of the model, Levenberg-Marquardt optimization algorithm was used. After many training attempts with various numbers of hidden neurons, the structure with 22 hidden neurons showed the best performance. R values for the test set were 0.986, 0.988 for validation, and 0.992 for training, according to regression analysis. Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) values remained consistently low across all normalized datasets, ranging from 0.0128 to 0.0148 for MAE and 0.0178 to 0.0238 for RMSE. These results underscore the model's robust predictive capabilities.