High Coefficient Of Determination Research Articles

Combination of dynamic TOPMODEL and machine learning techniques to improve runoff prediction

AbstractTOPMODEL has been widely employed in hydrology research, undergoing continuous modifications to broaden its practical applicability and enhance its simulation accuracy. To encompass spatial discretization, diffusion‐wave characteristics, depth‐dependent flow velocity, and flux estimation in the unsaturated zone, a generalized dynamic TOPMODEL is developed by introducing a greater number of physical parameters. The present study aims to evaluate the optimal combination of these parameters within the dynamic TOPMODEL framework using machine learning techniques to improve the accuracy of runoff predictions and bolster the model's reliability. An innovative training method is suggested to elevate the model's performance by integrating the Long Short‐Term Memory (LSTM) algorithm and a topological classification, which relies on the evolving spatial distribution of runoff conditions during floods. The research findings show that the proposed methodology achieves the lowest mean relative error (MRE) at 0.106, the highest Pearson correlation coefficient (PC) at 0.938, and the highest coefficient of determination (R2) at 0.906 among the three dynamic TOPMODEL types adopted in this study. The effective implementation of a case study in a river basin showcases the feasibility of the proposed method in conjunction with dynamic TOPMODEL and underscores the importance of employing the suggested training procedure.

GAIA: A New Formula for Reference Evapotranspiration

Estimating evapotranspiration is crucial for irrigation and agricultural applications. Although the FAO-56 Penman–Monteith method is highly accurate under conditions of abundant data, its extensive requirements limit its practical application. In the Mediterranean region, most empirical formulas used to estimate evapotranspiration are temperature-based and require calibration to be effective. The current study aims to introduce a novel formula to determine reference evapotranspiration using temperature, relative humidity, and extraterrestrial radiation daily in the Mediterranean region and evaluate its performance. Multi-linear regression was applied to agrometeorological data from the California Irrigation Management Information System (CIMIS) to develop the ETo formula. The formula was then validated using data from 252 stations in four countries (Greece, Spain, Portugal, and Cyprus) over the growing period of six years (2018–2023). The GAIA formula consistently outperformed formulas with the same or fewer variables, including Berti and Ahooghalaandari, across all metrics. The largest differences were observed in RMSE and the index of agreement. There is a strong correlation between GAIA and the FAO-56 Penman–Monteith formula (coefficient of determination = 0.88). While the GAIA formula shows a high coefficient of determination, its performance is somewhat lower than that of Copais and Valiatzas, particularly in terms of Pearson correlation and the coefficient of determination. A key difference is that Copais and Valiatzas rely on incoming solar radiation, whereas GAIA uses extraterrestrial radiation. Relative humidity was found to be the most influential variable, accounting for over 71% of the variance in ETo. Effective evapotranspiration (ETo) calculation methodologies, especially in areas with limited agrometeorological data, can significantly enhance irrigation system efficiency and promote sustainable water management. The GAIA formula offers a cost-effective method for estimating reference evapotranspiration (ETo) during the growing season with enhanced accuracy, eliminating the need for expensive equipment. However, its primary limitation is that it is validated in the Mediterranean region and within a specific geographical latitude range.

Modeling the electrical conductivity relationship between saturated paste extract and 1:2.5 dilution in different soil textural classes

Regression models were developed to estimate the electrical conductivity of saturated paste extract (ECe) from the electrical conductivity of soil-water ratio (EC1:2.5) for different soil textural classes. ECe is a crucial parameter used to indicate the presence, type, and distribution of salinity in soils. However, determining ECe is demanding, time-consuming, requires considerable skill to accurately identify the correct soil saturation point, and is not routinely performed by soil testing laboratories. Many laboratories, instead, commonly measure the electrical conductivity of soil-water extracts at various dilutions, such as EC1:1, EC1:2.5, or EC1:5. In this study, 706 soil samples were collected from depths of 0 - 30 cm across three rice irrigation schemes to determine EC1:2.5, with 50% analyzed for ECe. ECe values were grouped based on soil textural classes. The results showed a strong linear relationship between EC1:2.5 and ECe values, with a high coefficient of determination (R² > 0.95). The Root Mean Square Error values were low (1.4 < RMSE), and the Mean Absolute Error values were similarly low (0.85 < MAE). Therefore, the regression models developed provide a practical means of estimating ECe for various soil textural classes, thereby enhancing soil salinity assessment and management strategies.

Machine Learning Modeling of Anchovy Waste Treatment Using Solar Drying

ABSTRACTThis study aims to valorize coproducts from the anchovy processing chain by obtaining compounds of interest through the implementation of environmentally friendly and energy‐efficient techniques. These methods, which also apply to other fresh anchovy waste coproducts, seek to minimize the environmental pollution associated with conventional systems. The investigation focused on the application of solar drying as a treatment of anchovy waste. The resulting data were employed to model the drying behavior of anchovy waste using five machine learning algorithms. A thermokinetic study was conducted under both natural and forced convection solar drying to establish the optimal conditions for drying and storing anchovy heads, which are a significant source of high‐quality proteins for human and animal nutrition. Drying kinetics were examined at three temperatures (60°C, 70°C, and 90°C) and two airflow rates (150 and 300 m3/h). The study identified air drying temperature as the most critical factor affecting the drying kinetics of anchovy wastes. Machine learning modeling of anchovy waste solar drying was conducted, and evaluated models were RNN, LSTM, GRU, LightGBM, and CatBoost. CatBoost demonstrated superior performance in predicting moisture content. It achieved the lowest Mean Squared Error of 1.1491e − 06, the lowest Mean Absolute Error of 0.0006265, and the highest coefficient of determination (R2) of 99.99%. The comparative analysis highlighted distinct differences in the predictive accuracy of the models, with CatBoost emerging as the most effective.

Optimized Extraction of Sargahydroquinoic Acid, Major Bioactive Substance, from Sargassum yezoense Using Response Surface Methodology

Sargahydroquinoic acid (SHQA), a bioactive compound found in certain Sargassum species, exhibits significant health benefits. This study optimized the extraction of SHQA from Sargassum yezoense using response surface methodology (RSM) and evaluated its antioxidant effects through in vitro and in vivo assays. A Box–Behnken design (BBD) was effectively employed to investigate the effects of incubation temperature, time, and ethanol concentration on SHQA yield, achieving a high coefficient of determination (R2 = 0.961). Analysis of variance (ANOVA) validated the model’s reliability (F = 13.86, p = 0.005) and highlighted ethanol concentration as a highly significant factor (p < 0.001). Optimal extraction conditions were identified as 52.8 °C, 8.3 h, and 74.1% ethanol. The SHQA-maximized extract (SME) contained 67.8 ± 0.6 mg SHQA/g and 25.00 ± 1.01 mg phloroglucinol equivalent/g. SME exhibited antioxidant capacity of 26.45 ± 0.66 mg and 28.74 ± 2.30 mg vitamin C equivalent/g in ABTS and DPPH assays, respectively, and 0.29 ± 0.02 mM FeSO4 equivalent/g in the FRAP assay. Additionally, SME at 50 µg/mL and SHQA at 1 µg/mL inhibited reactive oxygen species (ROS) generation in an H2O2-induced zebrafish model. This study presents the first optimization of SHQA extraction using RSM and demonstrates SHQA’s ROS inhibition in a zebrafish model.

Design and synthesis of a new recyclable nanohydrogel based on chitosan for Deltamethrin removal from aqueous solutions: Optimization and modeling by RSM-ANN.

In this study, a new magnetic biocompatible hydrogel was synthesized as an adsorbent for Deltamethrin pesticide removal. The optimal conditions and adsorption process of Deltamethrin by chitosan/polyacrylic acid/Fe3O4 nanocomposite hydrogel was studied by Response Surface Methodology by Central Composite Design (RSM-CCD) and Artificial Neural Network (ANN). This adsorbents were synthesized, and then characterized and investigated using FT-IR, XRD, FE-SEM, EDX, Map, VSM, and TGA methods. The results of these analyses showed that the nanocomposite hydrogel was well synthesized and has the ability to adsorb the Deltamethrin pesticide. The results obtained through analysis using response surface methodology showed that the maximum amount of adsorption was 99.79% at 26°C, while pH, initial concentration, contact time, and adsorbent dose were 7, 22ppm, 90min, and 1.3g/L respectively. Comparison between results obtained from CCD modeling and artificial neural network proved that both methods had high ability to predict the adsorption process but the CCD method had higher coefficient of determination and lower error. The equilibrium and kinetic study of the process showed that the Toth isotherm model, pseudo-second-order is all suitable for expressing the adsorption process. In addition, the adsorption mechanism followed double-exponential model that combines external and internal diffusions. Results of Thermodynamic study suggested that the Deltamethrin adsorption on CS/PAA/Fe3O4 was a spontaneous and exothermic process. The results of the equilibrium process study revealed that the adsorption process was physical and desirable, therefore, the adsorption-desorption process was performed which showed that the composite was reusable up to 10cycles.

Monthly and annual rainfall erosivity in Poland: An empirical model including winter snowfall effect

Study regionPoland Study focusThis study proposes an empirical model to assess rainfall erosivity in Poland, with a particular focus on incorporating the influence of winter snowfall on soil erosivity. Utilizing a dataset collected from 225 meteorological stations across Poland, the research aims to refine both monthly and annual rainfall erosivity assessments. An important feature of this model is the integration of daily minimum temperatures to adjust erosivity calculations during the winter months, a period typically characterized by lower erosivity due to snowfall, which is less erosive compared to rainfall. New hydrological insights for the regionDuring the calibration phase of annual rainfall erosivity, the model achieved a high coefficient of determination (R²) of 0.77. Validation results were similar, with an R² of 0.72. Further, the model was assessed for its seasonal performance, revealing significant variations in effectiveness across the year. For example, during the spring and autumn months, the model showed good predictive capabilities with R² values of 0.77 and 0.68, respectively. For winter and summer the R² value were slightly lower, reaching 0.64 and 0.61, respectively. Overall, the model offers a useful tool for predicting erosivity with improved accuracy by accounting for snowfall effects, which can aid in enhancing soil erosion assessments and supporting more informed soil conservation strategies.

Factors influencing elderly coping with social challenges in Thailand's aging society

Thailand is facing a rapidly aging population, presenting numerous social challenges for the elderly that necessitate effective coping mechanisms to sustain their well-being. This study investigates factors influencing their ability to cope with aging-related challenges and the mechanisms driving their adaptation. A total of 330 participants aged 60 years and older were surveyed. Data analysis involved descriptive statistics, confirmatory factor analysis, structural equation modeling with AMOS, and path analysis. Qualitative data were collected through in-depth interviews and focus group discussions, with participants divided into three age groups: 60-69, 70-79, and 80+. Content analysis was applied to the qualitative data. Findings revealed that cooperation and societal support had the highest factor loading (0.809), while technological proficiency had the lowest (0.513). The structural equation model showed a high coefficient of determination (R² = 85%) for adaptation. The results underscore that adaptability is a key factor in coping with aging challenges, enabling physical and mental adjustment. Enhancing technological skills and promoting government policies and sector collaboration are vital for improving quality of life and readiness to address Thailand’s transition to a super-aged society.

Fatigue failure testing of human hair: Weibull-analysis for constant strain experiments

Fatigue failure testing of materials is an important aspect of assessing their strength and resilience under long-term, oscillatory stresses and/or strains. This also applies to human hair. For this investigation, we decided to complement existing experience on cyclic tests at various levels of constant stress with those at various constant strains (4–30%). For the description and analysis of the data sets, we opted for a non-linear fit of the cumulative two-parameter Weibull distribution (CWD) to the survival data. This gives direct access to the numerical values of the parameters as well as to their standard errors (SE), as measures of precision. As relevant parameters, we identified the lifetime index ln(α) and the shape factor β. All fits showed very high coefficients of determination and normally distributed residuals. Accordingly, precision of the parameter values is very high. It only starts to drop for high constant strains, when significant grouping of data starts to occur. ln(α) drops and β increases both exponentially with strain. β exceeds the value of unity (β ≥ 1) at a strain of 4.3%, indicating a fundamental change of failure mode. The cross-over of the theoretical curves for ln(α) and β occurs around 45% strain, which coincides with the break strain for conventional tensile testing. This agreement supports the validity of our approach and suggests a more than just empirical nature of the CWD-function for modelling the fatigue failure data of human hair.

Wave-induced motion prediction of a deepwater floating offshore wind turbine platform based on Bi-LSTM

The prediction of platform motions induced by irregular waves is of paramount importance for the safe monitoring and anticipatory control of a floating offshore wind turbine (FOWT) installed in deep sea areas. Within this study, a neural network method utilizing the Bi-directional Long Short Term Memory (Bi-LSTM) model is introduced to predict three representative motions of the OO-Star semi-submersible platform, designed for 10 MW wind turbines, with a focus on model hyperparameter optimization and the influence of noise. The neural network considers the high characteristic correlation between the wave conditions as input and platform motions as output to construct the Bi-LSTM model. Through the comparative analysis, the neurons, layers and optimizers were optimized to confirm the best configuration for the Bi-LSTM model structure. To verify the feasibility of the Bi-LSTM model, the platform motions under the wave configurations that are different from those used in training and validation are tested. With the help of the neural network, reliable dynamic responses up to 10 s can be predicted with a high determination coefficient of over 90% based on 30s-long historical motion data and the future wave kinematics that can be obtained by ocean data measurement devices. Furthermore, the model still maintains a reasonable accuracy even with the high noisy disturbance, indicating that the proposed model's ability to filter uncertainty factors. This research is anticipated to support to the development of structural condition monitoring and anticipatory control of floating offshore wind turbines.

Harnessing carbon-based adsorbents for poly- and perfluorinated substance removal: A comprehensive review

Poly- and perfluoroalkyl substances (PFAS) are synthetic, highly fluorinated chemicals commonly used in products like water-resistant materials, personal care items, non-stick cookware, and cleaning agents. Due to the strength of their CF bonds, PFAS are extremely persistent in the environment, creating significant challenges for pollution control. Traditional degradation methods often prove ineffective, making adsorption a promising alternative for PFAS removal from contaminated water. Carbon-based adsorbents, valued for their sustainability and efficiency, are particularly effective for this purpose. This review provides a detailed analysis on PFAS properties and their removal using various carbon-based adsorbents, including activated carbon, biochar, graphene oxide, carbon nanotubes, and magnetic composites. Adsorption capacities of these materials were seen vary significantly, from as low as 10.7 mg/g for Fe3O4-hybrid biochar, to as high as 713.85 mg/g for functionalized graphene-based adsorbent, highlighting the superior adsorption efficiency of modified graphene adsorbents compared to traditional biochar and activated carbons. Most of these adsorbents fit the Langmuir isotherm model and pseudo-second-order kinetic model with high coefficient of determination, indicating efficient monolayer adsorption and strong interaction with PFAS. The review also explores different adsorption mechanisms, the effects of process parameters on adsorption efficiency, strategies for adsorbent regeneration, and concludes with a bibliometric analysis that highlights key research gaps. These insights are intended to guide future advancements in the development of effective, scalable PFAS remediation technologies using carbon-based adsorbents.

Comparative Quantitative and Discriminant Analysis of Wheat Flour with Different Levels of Chemical Azodicarbonamide Using NIR Spectroscopy and Hyperspectral Imaging.

This study investigated and comprehensively compared the performance of spectra (950-1660 nm) acquired respectively from NIR and HSI in the rapid and non-destructive quantification of azodicarbonamide (ADA) content (0-100 mg/kg) in WF and simultaneously identified WF containing excessive ADA (>45 mg/kg). The raw spectra were preprocessed using 14 methods and then mined by the partial least squares (PLS) algorithm to fit ADA levels using different numbers of WF samples for training and validation in five datasets (NTraining/Validation = 189/21, 168/42, 147/63, 126/84, 105/105), yielding better abilities of NIR Savitzky-Golay 1st derivative (SG1D) spectra-based PLS models and raw HSI spectra-based PLS models in quantifying ADA with higher determination coefficients and lower root-mean-square errors in validation (R2V & RMSEV), as well as establishing 100% accuracy in PLS discriminant analysis (PLS-DA) models for identifying excessive ADA-contained WF in each dataset. Twenty-four wavelengths selected from a NIR SG1D spectra in a 168/42 dataset and 23 from a raw HSI spectra in a 147/63 dataset allowed for the better performance of quantitative models in ADA determination with higher R2V and RMSEV in validation (R2V > 0.98, RMSEV < 3.87 mg/kg) and for discriminant models in WF classification with 100% accuracy. In summary, NIR technology may be sufficient if visualization is not required.

Application of novel deep neural network on prediction of compressive strength of fly ash based concrete

ABSTRACT Fly ash (FA)-based high-strength concrete (HSC) has attracted significant interest due to its potential to substitute Portland cement, offering both environmental benefits and improved performance. However, the design of FA-HSC is challenging, as key factors such as fly ash percentage, water content, and superplasticizer dosage have a complex influence on compressive strength. This study aims to develop an efficient predictive tool for FA-HSC mix design, using artificial intelligence (AI) models to address the inherent variability and uncertainty in these parameters. Six AI models, including a Deep Neural Network (DNN), were employed to analyse the relationships between mix design variables and compressive strength. The DNN model, in particular, demonstrated superior performance compared to the other models, with a high coefficient of determination (R2 = 0.89), variance accounted for (VAF = 88.3%), root mean square error (RMSE = 0.06), and residual standard error (RSR = 0.31). These results indicate that the DNN model can provide reliable predictions of compressive strength, offering a more efficient alternative to traditional trial-and-error methods. The AI-based approach can save both time and material costs while optimising performance. Overall, this AI-driven model contributes to the advancement of sustainable concrete technology by enabling more precise and resource-efficient mix designs for FA-based high-strength concrete.

Temperature-dependent compressive strength modeling of geopolymer blocks utilizing glass powder and steel slag

This article investigates the development of geopolymers as a modern, environmentally sustainable binder with ceramic-like properties, offering exceptional thermal and fire-resistant characteristics. The study primarily utilized fly ash (FA) in combination with glass powder (GP) and steel slag (SS). The SS content varied between 30 % and 40 %, while the molarity of NaOH was set at 10 M, 12 M, and 14 M. Based on these variables, a total of eighteen mixes incorporating GP and SS were formulated. The samples were subjected to elevated temperatures of 200 °C, 400 °C, 600 °C, and 800 °C, after which their compressive strengthswere measured. To better understand the material formation, analyses were conducted by using scanning electron microscopy, energy dispersive X-ray spectroscopy, X-ray diffraction, Fourier transform infrared spectroscopy, and thermogravimetry differential thermal analysis. The investigation examined the influence of oxide ratios (Na/Si, Si/Al, H2O/Na2O, and Na/Al) on the compressive strength at elevated temperatures. Additionally, the research sought to develop a predictive model, elucidating the relationship between these oxide ratios and the compressive strength of geopolymers. To achieve this, ten machine learning techniques were applied, revealing the complex connection between oxide ratios and the strength properties of geopolymers. The support vector regressor (SVR) model outperformed other regression and boosting models, obtaining a high coefficient of determination (R2) value of 0.95, indicating superior predictive accuracy. The reduced error levels and high R2 values highlighted the enhanced performance of the SVR model. A sensitivity analysis was done to understand the contributions of each parameter to the outcome predictions further. Employing machine learning techniques to predict the compressive strength of geopolymer blocks under various elevated temperature conditions improves predictive accuracy and optimizes resource utilization, leading to significant time savings.

Mathematical Modeling and Optimization of Enzyme‐Based Amperometric Screen‐Printed Biosensors Using Horseradish Peroxidase as a Model

AbstractIn this study, a mathematical model was formulated to characterize the functioning of third‐generation enzyme‐based amperometric biosensors. This study utilized a horseradish peroxidase (HRP)‐based, screen‐printed biosensor. The model parameters were determined using the Environment for Modeling Simulation and Optimization (EMSO), validated with experimental data, and exhibited high determination coefficients, indicating the accuracy of the models in capturing biosensor behavior. Simulations based on this model emphasize diffusion as a limiting factor. Further sensitivity analysis was conducted using simulated response surfaces and biosensor performance optimization, which confirmed that higher concentrations of HRP in the thinnest polymeric film resulted in enhanced biosensor sensitivity.

From dishwasher to river: how to adapt a low-cost turbidimeter for water quality monitoring.

This study presents the process of design and development of a low-cost turbidimeter for monitoring water quality, facilitating rigorous spatial-temporal variability analysis within large-scale hydrological systems. We propose a low-cost optical turbidimeter, modifying the existent SEN0189 turbidity sensor, Arduino boards, and additional sensors for temperature compensation. We compared a low-cost system with high-tech sensors, modifying the original low-cost SEN0189 probe for enhanced environmental performance. The three-step methodological framework involved prototype development, compensation for environmental factors, and preparation for future field deployment. Calibration equations with a high coefficient of determination and a temperature correction equation were established. We made adaptations to overcome field deployment challenges, including a 3-D printed sensor case, defining the relationship between measurement uncertainty and energy consumption, and specifying field installation guidelines. In summary, this study presents a comprehensive approach to a low-cost optical turbidity system, demonstrating its potential for accurate and affordable field deployment. We aim to address the critical need for sustainable inland water management tools, making this system a valuable contribution to environmental monitoring practices. We also aim to inspire similar development of open-source monitoring systems within our community.

Sonophotocatalytic degradation of rhodamine B dye using Zr doped ZnO nanoparticles: Kinetic study and toxicity assessment

This study investigates sonophotocatalysis for the oxidation of Rhodamine B (RhB) dye using Zr doped ZnO nanoparticles. The synthesized nanoparticles were characterized by scanning electron microscope, Fourier-transform infrared, X-ray diffraction, Brunauer Emmett-Teller, thermogravimetric analysis, and photoluminescence. The hybrid sonophotocatalysis system achieved 98.73 ± 1.25% degradation of RhB under the conditions: RhB dye concentration = 10 ppm, 4 wt% of Zr doped ZnO = 0.1 g, reaction temperature = 30 °C, time = 60 min, pH = 7. A kinetic model was developed to predict the efficiency of RhB degradation through intrinsic elementary chemical reactions. The high coefficient of determination (R2 = 0.96) indicates the model's prediction accuracy in describing the degradation kinetics of RhB within sonophotocatalysis system. The electrical energy per order (EEO) analysis demonstrated that US + UVC + 4 wt% of Zr doped ZnO significantly reduced EEO (1055 kWh m−3 order−1). The synergy index for this combination was approximately 1.60, demonstrating a significant synergistic effect. Density Functional Theory (DFT) studies were utilized to propose the most probable degradation pathway, identifying reactive sites and potential byproducts. The simulations revealed the central carbon atom (1C site) as highly susceptible to radical attacks, indicating potential decolorization pathways such as N-de-ethylation, chromophore cleavage, and ring opening. Toxicity assessments of both the parent dye and its intermediates were conducted using ECOSAR (Ecological Structure Activity Relationship) to evaluate their ecological impacts. The combination of experimental results and kinetic modeling and simulations offers a deep understanding of RhB degradation complexities, driving advancements in sustainable water treatment technologies.

Development of FTIR-ATR spectra and PLS regression combination model for discrimination of pure and adulterated acacia honey

A Fourier Transform Infrared Spectroscopy and attenuated total reflectance (FTIR-ATR)-based chemometric model was evaluated for the rapid identification and estimation of adulterants in honey. Two types of honey, bee honey and stingless bee honey, were adulterated with acid adulterants (acetic acid, citric acid, and tamarind extract) at different concentrations of 1%, 3%, 5%, and 7% and sugar adulterants (liquid corn syrup, cane sugar, palm sugar, and inverted sugar) at different concentrations of 1%, 2%, and 3%. The physicochemical properties (Brix, moisture content, pH, and free acidity) and sugar contents (glucose, fructose, and sucrose) of each type of honey and their adulterated samples were determined before FTIR analysis. For all the samples, FTIR spectra were acquired in the mid-infrared range (400-3600 cm−1) and the spectra obtained were subjected to partial least squares (PLS) analysis to develop the model for the determination of adulterants in honey samples. The PLS model produced high coefficient of determination (R2) for bee honey adulterated with acetic acid (0.95), citric acid (0.98), tamarind extract (0.97), and stingless bee honey adulterated with liquid corn syrup (0.99), inverted sugar (0.91), cane sugar (0.86), and palm sugar (0.77). A recognition model was developed for mixture detection and verified through the physicochemical analysis. The combination of FTIR-ATR spectra and the PLS regression model can be a fast, reliable, nondestructive method to detect honey adulteration.

Optimising the production of PLGA nanoparticles by combining design of experiment and machine learning

Poly(lactic-co-glycolic acid) (PLGA) is a widely used biodegradable polymer in drug delivery and nanoparticle (NP) formulation due to its controlled drug release properties and safety profiles. Among the methods available for NP production, nanoprecipitation is distinguished by its simplicity and scalability. However, it requires careful optimisation to achieve the desired NP characteristics, making the process potentially lengthy and costly. This study aimed to assess and compare the predictive performance of Design of Experiments (DOE) and Machine Learning (ML) models for the optimisation of PLGA nanoparticle size and zeta potential produced by nanoprecipitation. Various ML methods were employed to predict particle size, with Extreme Gradient Boosting (XGBoost) identified as the best performing. The key finding is that integrating ML with DOE provides deeper insights into the dataset than either method alone. While ML outperformed DOE in predictive performance, as evidenced by lower root mean squared error values and higher coefficients of determination, both methods struggled to accurately predict zeta potential, generating models with high errors. However, ML proved more effective in identifying the parameters that most significantly influence NP size, even with a smaller DOE dataset. Combining DOE datasets with ML for parameter importance was particularly advantageous in situations where data is limited, offering superior predictive power and the potential to streamline experimental design and optimisation. These results suggest that the synergistic use of ML and DOE can lead to more robust feature analysis and improved optimisation outcomes, particularly for NP size. This integrated approach can enhance the accuracy of predictions and supports more efficient experimental design, streamlining nanoparticle production processes, especially under resource-constrained conditions.

Use Artificial Neural Networks to Predict Seepage in the Hilla Canal Regulator

The variation in the seepage under hydraulic structures significantly impacts their stability and effective water management, especially considering recent water scarcity challenges. This paper aims to calculate seepage and investigate the hydraulic performance of the Al-Hilla canal regulator's foundation. The methodology involves constructing a model (SEEP/W) and comparing its results with one-site piezometer readings. Using a mixed-methods approach, this study integrates software modelling and statistical analysis techniques. This study integrates software modelling and statistical analysis techniques. The Geo-studio program facilities modelling of seepage flow, while JASP software is used for statistical analysis and predictive equation development. The methodology consists of data collection, Geo-Studio modelling, JASP analysis, and validation of equation accuracy. After verifying the models’ efficiency, the data for the seepage equation was established under various upstream and downstream conditions, incorporating artificial intelligence algorithms. This data was analyzed to drive a predictive equation for seepage with a high coefficient of determination (R2) OF 97%. Additionally, another equation was formulated to determine the total water pressure head, achieving an R2 value of 95 %. These equations are invaluable tools for predicting the total water pressure head and seepage and enhancing the management of hydraulic structure.