Search

In recent years, machine learning (ML) algorithms have gained substantial recognition for ecological modeling across various temporal and spatial scales. However, little evaluation has been conducted for the prediction of soil organic carbon (SOC) on small data sets commonly inherent to long-term soil ecological research. In this context, the performance of ML algorithms for SOC prediction has never been tested against traditional process-based modeling approaches. Here, we compare ML algorithms, calibrated and uncalibrated process-based models as well as multiple ensembles on their performance in predicting SOC using data from five long-term experimental sites (comprising 256 independent data points) in Austria. Using all available data, the ML-based approaches using Random forest and Support vector machines with a polynomial kernel were superior to all process-based models. However, the ML algorithms performed similar or worse when the number of training samples was reduced or when a leave-one-site-out cross validation was applied. This emphasizes that the performance of ML algorithms is strongly dependent on the data-size related quality of learning information following the well-known curse of dimensionality phenomenon, while the accuracy of process-based models significantly relies on proper calibration and combination of different modeling approaches. Our study thus suggests a superiority of ML-based SOC prediction at scales where larger datasets are available, while process-based models are superior tools when targeting the exploration of underlying biophysical and biochemical mechanisms of SOC dynamics in soils. Therefore, we recommend applying ensembles of ML algorithms with process-based models to combine advantages inherent to both approaches. • We evaluated process-based (PB) and machine-learning (ML) models for SOC prediction. • Random forest & support vector machine predicted SOC superior to PB models. • Reduced data availability and quality significantly weakened ML model performance. • ML-assisted PB model ensembles strongly improved model performance. [ABSTRACT FROM AUTHOR]

• FFS maps were produced using CTREE, KNN, SVM, RF, GBM and XGBoost algorithms. • The performance of six machine learning techniques was evaluated. • This study will be the first to use the CTREE algorithm for FFS mapping. • XGBoost gave the best results compared to other models. • The most important factor for producing FFS maps in XGBoost models was annual temperature. Antalya is one of the provinces with the highest number of forest fires in Türkiye. In 2021, 278 forest fires occurred within the administrative boundaries of Antalya Regional Directorate of Forestry. The main objective of this study is to produce forest fire susceptibility (FFS) maps of Antalya province using machine learning (ML) models. In addition to forest fire inventory data, 16 factors, including topographic, environmental, meteorological, and human-driven, were used in the study. Inventory data included 2166 fire ignition points from the General Directorate of Forestry. 70 % of the inventory dataset was used to train the ML models and 30 % to validate the models. Overall accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC-ROC) approaches were considered as validation metrics. FFS maps of Antalya were produced using stand-alone ML algorithms, K-Nearest Neighbors, and Support Vector Machines, as well as tree-based Conditional Inference Trees (CTREE), Random Forest (RF), Gradient Boosting Machines (GBM), and Extreme Gradient Boosting (XGBoost) algorithms. To the best of our knowledge, this is the first study using the CTREE algorithm for forest fire susceptibility mapping. Therefore, this study is important for the related literature. The validation results revealed that the XGBoost model outperformed other models. It is thought that the FFS map produced using the XGBoost model will guide forest engineers, wildland firefighting teams, and firefighters to minimize damage and control forest fires. [ABSTRACT FROM AUTHOR]

The estimation of material performance plays a crucial role in practical life, enabling the rational allocation of time and resources while enhancing the practical application of materials. Therefore, this study investigates the analytical and predictive capabilities of five machine learning (ML) algorithms (the Random Forest algorithm – RF, the Adaboost algorithm – AB, the Gradient Boosting algorithm – GB, the Extra Trees algorithm – ET and linear models – LM) for comprehensive performance parameters of wood ceramics (ammonia nitrogen adsorption quantity - Q , open porosity - P , burn-off , and hardness). In the analysis of model prediction capabilities, five key statistical parameters (Root Mean Square Error - RMSE , Mean Square Error - MSE , Coefficient of Correlation - R , Determination Coefficient - R 2 , and Mean Absolute Relative Error - MARE) were calculated. The results indicate the following: (1) Among various ML models, the GB model exhibits the most superior performance with an R 2 ≥ 0.9594. (2) However, in predicting the four performance parameters of wood ceramics, the AB model with an R 2 ranging from 0.0078 to 0.45, indicating notably poor predictive capabilities. Despite integrating the Support Vector Regression (SVR) module, no enhancement in predictive accuracy was observed. (3) While forecasting the four performance parameters of wood ceramics, the LM model demonstrates predictive accuracy relatively akin to that of the RF and ET models among the ML models. (4) Though the feature importance scores vary across distinct input variable models, there is a consistent trend of change. (5) Variables B and D exhibit some level of correlation with other variables. In summary, the results of this study suggest that, for the analysis and prediction of wood ceramic performance, the GB model among the five regression models demonstrates outstanding simulation performance. [ABSTRACT FROM AUTHOR]

• This study examined the genes related to PD pathogenesis by tissues and blood samples datasets. • WGCNA and protein–protein interaction (PPI) network were used to pinpoint hub genes. • Elevation of FOXO3 , HIST2H2BE , and HDAC1 expression associated with PD pathogenesis. Parkinson's disease (PD) represents a multifaceted neurological disorder whose genetic underpinnings warrant comprehensive investigation. This study focuses on identifying genes integral to PD pathogenesis and evaluating their diagnostic potential. Initially, we screened for differentially expressed genes (DEGs) between PD and control brain tissues within a dataset comprising larger number of specimens. Subsequently, these DEGs were subjected to weighted gene co-expression network analysis (WGCNA) to discern relevant gene modules. Notably, the yellow module exhibited a significant correlation with PD pathogenesis. Hence, we conducted a detailed examination of the yellow module genes using a cytoscope-based approach to construct a protein–protein interaction (PPI) network, which facilitated the identification of central hub genes implicated in PD pathogenesis. Employing two machine learning techniques, including XGBoost and LASSO algorithms, along with logistic regression analysis, we refined our search to three pertinent hub genes: FOXO3 , HIST2H2BE , and HDAC1 , all of which demonstrated a substantial association with PD pathogenesis. To corroborate our findings, we analyzed two PD blood datasets and clinical plasma samples, confirming the elevated expression levels of these genes in PD patients. The association of the genes with PD, as reflected by the area under the curve (AUC) values for FOXO3 , HIST2H2BE , and HDAC1 , were moderate for each gene. Collectively, this research substantiates the heightened expression of FOXO3 , HIST2H2BE , and HDAC1 in both PD brain and blood samples, underscoring their pivotal contribution to the pathogenesis of PD. [ABSTRACT FROM AUTHOR]

Machine learning applications in the medical field has proliferated in the past decade. One of the main usages of machine learning in medical field is the detection of diseases from text and image data. In this work, a new hybrid multi-layer algorithm (MLHA) based on machine learning algorithms is proposed to predict diabetes. The algorithm consists of two layers. The first layer consists of three different machine learning algorithms; SVM, random forest and K-mean, that work in parallel on the dataset. The output of this layer is fed into the second layer that consists of only one algorithm; XGBoost, that is trained from the output data of the first layer. We trained the proposed model with two different datasets and its accuracy has been compared with three different common well-known algorithms; feed forward back propagation neural network, logistic regression and random forest. Our results show that the accuracy of MLHA reached 86.5% for the first dataset and approximately 84.6% for the second dataset. This reported accuracy outperformed all the classical models. [ABSTRACT FROM AUTHOR]

• Younger older people who participated in employment experienced better mental health. • The association between employment and mental health presents significant heterogeneity. • Gender, age, residence, marital status, educational attainments, and self-rated health all contribute to the heterogeneous association, but pension benefits are the dominant factor. • There was a non-linearity in the conditional average treatment effects across the distribution of pensions. This study aimed to examine associations between the employment of older people and mental health across demographic characteristics, socioeconomic conditions, and health status, with a focus on pensions. This study included 4,512 participants aged 60–69 from the CLASS in 2014. A multiple linear regression was conducted to investigate the association between employment and mental health. A causal forest model was applied to estimate the heterogeneous treatment effects. Employed individuals (n = 1,295) reported better mental health than their non-employed counterparts. This association displayed significant heterogeneity, primarily attributed to pensions. Those with lower pensions may be compelled to work due to financial reasons, thus offsetting the health-promotion effect of employment. Employment may benefit the mental health of older adults, which has a more significant marginal effect on those who are men, older, urban residents, without a spouse, below primary education, receiving more pensions, and less family and friend support. [ABSTRACT FROM AUTHOR]

VEGF is prototypic marker of neovascularization, repeatedly proposed as intrinsic characteristic of peri-implantitis. This study aimed to assess pattern of VEGF in peri-implantitis, its correlation with titanium particles (TPs) and capacity as respective biomarker. Pathological specificity of VEGF was assessed in peri-implant granulations using immunohistochemistry, periodontal granulations represented Ti-free positive controls. VEGF was correlated to TPs, identified using scanning electron microscopy coupled with dispersive x-ray spectrometry. Diagnostic accuracy, sensitivity and specificity of VEGF were estimated in PICF specimens from peri-implantitis, peri-implant mucositis (PIM) and healthy peri-implant tissues (HI) using machine learning algorithms. Peri-implantitis exhibited rich neovascular network with expressed density in contact zones toward neutrophil infiltrates without specific pattern variations around TPs, identified in all peri-implantitis specimens (mean particle size 8.9 ± 24.8 µm2; Ti-mass (%) 0.380 ± 0.163). VEGF was significantly more expressed in peri-implantitis (47,065 ± 24.2) compared to periodontitis (31,14 ± 9.15), and positively correlated with its soluble concentrations in PICF (p = 0.01). VEGF was positively correlated to all clinical endpoints and significantly increased in peri-implantitis compared to both PIM and HI, but despite high specificity (96%), its overall diagnostic capacity was average. Two patient clusters were identified in peri-implantitis, one with 8-fold higher VEGF values compared to HI, and second with lower values comparable to PIM. VEGF accurately reflects neovascularization in peri-implantitis that was expressed in contact zones toward implant surface without specific histopathological patter variation around TPs. VEGF answered requests for biomarker of peri-implantitis but further research is necessary to decrypt its exact underlying cause. [Display omitted] • Neovascularization was expressed in contact zones toward implant surface. • VEGF is highly specific to peri-implantitis and accurately reflects neovascularization. • VEGF could not be directly associated to titanium particles. • In-depth research is required to decipher the specific VEGF pattern in peri-implantitis. [ABSTRACT FROM AUTHOR]

The color of road signs is a critical aspect of road safety, as it helps drivers quickly and accurately identify and respond to these signs. Properly colored road signs improve visibility during the day and make it easier for drivers to make informed decisions while driving. In order to ensure the safety and efficiency of road traffic, it is essential to maintain the appropriate color level of road signs. The objective of this study was to analyze the color status and daylight chromaticity of in-use road signs using supervised machine learning models, and to explore the correlation between road sign's age and daylight chromaticity. Three algorithms were employed: Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN). The data used in this study was collected from road signs that were in-use on roads in Sweden. The study employed classification models to assess the color status (accepted or rejected) of the road signs based on minimum acceptable color levels according to standards, and regression models to predict the daylight chromaticity values. The correlation between road sign's age and daylight chromaticity was explored through regression analysis. Daylight chromaticity describes the color quality of road signs in daylight, that is expressed in terms of X and Y chromaticity coordinates. The study revealed a linear relationship between the road sign's age and daylight chromaticity for blue, green, red, and white sheeting, but not for yellow. The lifespan of red signs was estimated to be around 12 years, much shorter than the estimated lifespans of yellow, green, blue, and white sheeting, which are 35, 42, 45, and 75 years, respectively. The supervised machine learning models successfully assessed the color status of the road signs and predicted the daylight chromaticity values using the three algorithms. The results of this study showed that the ANN classification and ANN regression models achieved high accuracy of 81% and R2 of 97%, respectively. The RF and SVM models also performed well, with accuracy values of 74% and 79% and R2 ranging from 59% to 92%. The findings demonstrate the potential of machine learning to effectively predict the status and daylight chromaticity of road signs and their impact on road safety in the Swedish context. • The study used supervised machine learning models to analyze road sign color and daylight chromaticity in Sweden. • The study found a linear relationship between age and daylight chromaticity for blue, green, red, and white sheeting but not for yellow. • The estimated lifespan of red signs is 12 years, while the estimated lifespans of yellow, green, blue, and white sheeting are 35, 42, 45, and 75 years, respectively. • The study found that the relationship between age and daylight chromaticity is affected by color, class, GPS coordinates, direction, and coefficient of retroreflection. • Machine learning models achieved high accuracy in predicting color status and daylight chromaticity, ranging from 74 to 81% for classification and R2 ranging from 59 to 97% for regression. [ABSTRACT FROM AUTHOR]

Unlike many previous studies that often focus on optimizing energy efficiency for buildings when detailed design drawings are available, this paper introduces a newly integrated model for energy-efficient building envelope design in the early stages (when detailed design drawings are not yet available). The newly developed model includes three main components: a simulation model, a predictive model, and an optimization model. The simulation model simulates the building's energy performance, considering different values for various envelope parameters. The predictive model employs machine learning algorithms, including RF, ANN, DNN, SVM, GENLIN, and GB (in which GB has been identified as the most suitable algorithm), boasting a very high R 2 (0.994) to assess energy consumption. The optimization model which uses AI optimization algorithms (such as NSGA II, DSE, and MOPSO) integrates the machine learning predictive model into the evaluation function during the evolutionary process, efficiently searching for Pareto-optimal building envelope solutions. Results show simultaneous savings in cost and energy, with savings of 7.52 % in cost and 8.48 % in energy, or 21.17 % in cost and 0.4 % in energy, for a case study in Vietnam. This model establishes a foundation by providing design solutions for stakeholders to assess, and can incorporate additional objectives at later stages. [ABSTRACT FROM AUTHOR]

In the industrial production process, it is important to improve the productivity of blast furnaces in response to the worldwide call for energy saving and emission reduction. In this study, a data-driven method using computational fluid dynamics (CFD) is developed for the rapid prediction of nitrate decomposition in a blast furnace. The method includes the creation of a CFD database, for which nozzle arrangement patterns, relative distances, spray cone angles and temperatures are simulated and collected. Support vector regression (SVR) hyperparameters are optimized using genetic algorithm (GA), particle swarm optimization (PSO), and gray wolf optimizer (GWO). Machine learning interpretability methods shapley additive explanations (SHAP) and partial dependence plots (PDP) are used to analyze the impact of features on prediction results. The results show that the hybrid GWO-SVR model has a better performance, with the degree of fit (R2) improving from 0.7752 to 0.9719. When the distance is taken around 0.58, increasing the temperature and cone angle is more favorable to improve the decomposition rate of the blast furnace. SHAP method shows that only the arrangement 2 and 4 have a positive effect on improving the decomposition rate. The response time for accurate prediction of nitrate decomposition rate is 259200 times shorter than that of industrial-scale CFD simulations. The method is also used to determine the best combination of solutions. This cross-disciplinary approach provides a time-saving and cost-effective tool for improving the response of blast furnace productivity to changes in nozzle arrangement. [ABSTRACT FROM AUTHOR]

In the field of computer network communication, there are various types of data, and these data have strong commercial value. Therefore, it is particularly important to apply them to relevant security monitoring. This article analyzes and studies the problems in remote detection of computer networks using machine learning algorithms. Firstly, three key technologies based on classifiers, feature extraction, and association rule recognition were introduced. Secondly, how to use machine learning methods to evaluate the performance indicators and operational effectiveness data statistics of computer network security monitoring systems was explained. Finally, this article tests the monitoring technology of the system, and the test results show that the security of the technology for data monitoring can reach over 94%. [ABSTRACT FROM AUTHOR]

How to deal with damage to the power grid caused by natural disasters has become one of the main research focus of disaster prevention and mitigation. This paper collates two main approaches to distribution network damage prediction and assessment under typhoon disasters: based on physical model-driven and data model-driven. Research based on physical models is divided into three parts by considering real-time wind speed correction, meteorological forecast information, and non-meteorological information. The field of data-driven models includes single-algorithm models and multi-algorithm models. These methods are known for their high accuracy, but have a strong dependence on data. Finally, the paper explores the potential benefits of integrating physical and data model-driven approaches, as well as implementing additional defensive measures based on predictive results. [ABSTRACT FROM AUTHOR]

Issues of providing mental health support to people with emerging or current mental health disorders are becoming a significant concern throughout the world. One of the biggest effects of digital psychiatry during COVID-19 is its capacity for early identification and forecasting of a person's mental health decline resulting in chronic mental health issues. Therefore, through this study aims at addressing the hological problems by identifying people who are more likely to acquire mental health issues induced by COVID-19 epidemic. To achieve this goal, this study includes 1) Rajyoga practitioners' perceptions of psychological effects, levels of anxiety, stress, and depression are compared to those of the non practitioners 2) Predictions of mental health disorders such as stress, anxiety and depression using machine learning algorithms using the online survey data collected from Rajyoga meditators and general the population. Decision tree, random forest, naive bayeBayespport vector machine and K nearest neighbor algorithms were used for the prediction as they have been shown to be more accurate for predicting psychological disorders. The support vector machine showed the highest accuracy among all other algorithms. The f1 score was also the highest for support vector machine. [ABSTRACT FROM AUTHOR]

The quest to attain net-zero emissions has increased the drive for more renewable energy potential from the co-gasification of biomass. The co-gasification of coconut shell and oil palm wastes is a potential technological route to produce hydrogen-rich syngas. However, the complexity of the gaseous-phase reaction often results in process uncertainties which could lead to energy and material wastage. Taking advantage of the data generated from the process, this study explores the performance of twelve machine learning algorithms built on the support vector machine (SVM), the Gaussian process regression (GPR), and the non-linear response quadratic model (NLRQM) using Sequential quadratic programming, and the Levenberg-Marquardt algorithms. The co-gasification of coconut shell and oil palm wastes blend catalyzed by Portland cement, dolomite, and limestone resulted in the maximum syngas production of 42 mol.%, 38 mol.%, 45 mol.%, respectively. The co-gasification process was modeled using SVM regression incorporated with linear, quadratic, and cubic kernel functions, GPR incorporated with rotational, squared, Matern 5/2, and exponential kernel functions, and non-linear response quadratic model (NLRQM) using sequential quadratic programming (SQP), and Levenberg-Marquardt (LM). The performance analysis of the models revealed that the SVM incorporated with linear kernel had the least performance with R2 in the range of 0.3–0.7. Whereas the best performance in terms of prediction of the syngas composition was obtained using the NLRQM algorithm with an inbuilt SQP and LM algorithms. The observed syngas composition was consistent with predicted values with R2 > 0.97 for the three catalyzed co-gasification processes. The low RMSE (<1) and MAE (<1) obtained from the models are indications of the robustness of the accurate prediction of the NLRQM-LM and NLRQM-SQP algorithms. The sensitivity analysis revealed that the co-gasification temperature, catalysts loading, and the blending amount play a significant role in the predicted syngas composition. However, the co-gasification temperature had the highest influence as indicated by the level of importance values. • Hydrogen-rich syngas production from co-gasification of biomass blend was investigated. • Twelve machine learning algorithms were employed to model the co-gasification process. • Support vector machine and Gaussian process regression performed moderately. • Non-linear response quadratic model embedded with inbuilt algorithms perform best. • Gasification temperature significantly influenced the predicted syngas composition. [ABSTRACT FROM AUTHOR]

High-performance alkali-activated concrete (HP-AAC) is acknowledged as a cementless and environmentally friendly material. It has recently received a substantial amount of interest not only due to the potential it has for being used instead of ordinary concrete but also owing to the concerns associated with climate change, sustainability, reduction of CO 2 emissions, and energy consumption. The characteristics and amounts of the ingredients used to produce HP-AAC influence its compressive strength. This study performs a comparative analysis based on machine learning (ML) algorithms to present an ensemble model capable of predicting the compressive strength of HP-AAC. This is in response to the development of sophisticated prediction approaches that seek to lower the cost of experimental tools and labor. An extensive framework including 538 experimental datasets with 18 input parameters are extracted. In addition, stacked ML (SM) models are developed to provide their best base estimator combination with the highest capability. The results show that stacked model (SM-5) with score of 14, and prediction accuracy of 98% following by the largest experiment-to-predicted ratio, provide the best estimations of compressive strength of HP-AAC, which has the lowest error values compare to other 18 ML models. Thereafter, a graphical user interface (GUI) is provided and validated by extra experimental tests for estimating the compressive strength, cost, and carbon emission of HP-AAC. Overall, the significance of the current study highlight the outstanding performance of developed stacked ML and GUI for predicting the compressive strength of HP-ACC, which contribute for the on-going research in this area. [Display omitted] [ABSTRACT FROM AUTHOR]

MicroRNAs (miRs) have emerged as promising biomarkers for early disease diagnosis and personalized treatment monitoring. However, their clinical utility has been hampered by technical limitations. Dynamic chemical labeling (DCL) based on capturing abasic PNA probes and reactive nucleobases, known as SMART bases, is a PCR-free approach that has proven very useful for the direct interrogation of circulating miRs. In this work, we expand the palette of tools available for the detection of DCL miR by synthesizing a new SMART nucleobase called SMART-C-Eu. This nucleobase contains a stable lanthanide cryptate. Using this SMART-C-Eu base and time-gated (TG) luminescence imaging, we successfully detect and quantify miR-122–5p in human serum samples. miR-122–5p is a well-known biomarker for drug-induced liver injury. Through a bead-counting analysis approach, statistical robustness is improved and miR-122–5p concentrations are detected in the nanomolar range. Furthermore, we extend this approach to multiplexed detection of three different miRs (miR-371a-3p, miR-451a-5p, and miR-122–5p) using spectral and temporal filtering. Importantly, we designed a user-independent multiplexed analysis using machine learning algorithms for automatic bead classification. Although the sensitivity of this technique must be further improved to detect miRs at lower concentrations, the method represents a significant advancement in miR analysis by combining ML segmentation using lifetime and intensity images. In addition, the technique offers multiplexing capabilities and the potential for automation, paving the way for more accurate and robust clinical applications in the future. [Display omitted] • DCL & SMART bases enable PCR-free miR biomarkers detection in serum. • New lanthanide luminophore SMART-C-Eu nucleobase introduced. • Successful quantification of miR-122–5p in serum using SMART-C-Eu & TG imaging. • Multiplexed detection of miRs achieved via lifetime and spectral filtering. • ML algorithms aid in user-independent multiplexed analysis. [ABSTRACT FROM AUTHOR]

The progression of gastric cancer involves a complex multi-stage process, with gastroscopy and biopsy being the standard procedures for diagnosing gastric diseases. This study introduces an innovative non-invasive approach to differentiate gastric disease stage using gastric fluid samples through machine-learning-assisted surface-enhanced Raman spectroscopy (SERS). This method effectively identifies different stages of gastric lesions. The XGBoost algorithm demonstrates the highest accuracy of 96.88% and 91.67%, respectively, in distinguishing chronic non-atrophic gastritis from intestinal metaplasia and different subtypes of gastritis (mild, moderate, and severe). Through blinded testing validation, the model can achieve more than 80% accuracy. These findings offer new possibilities for rapid, cost-effective, and minimally invasive diagnosis of gastric diseases. [Display omitted] [ABSTRACT FROM AUTHOR]

Osteosarcoma (OS), as the most common primary bone cancer, has a high invasiveness and metastatic potential, therefore, it has a poor prognosis. This study identified early diagnostic biomarkers using miRNA expression profiles associated with osteosarcoma metastasis. In the first step, we used RNA-seq and online microarray data from osteosarcoma tissues and cell lines to identify differentially expressed miRNAs. Then, using seven feature selection algorithms for ranking, the first-ranked miRNAs were selected as input for five machine learning systems. Using network analysis and machine learning algorithms, we developed new diagnostic models that successfully differentiated metastatic osteosarcoma from non-metastatic samples based on newly discovered miRNA signatures. The results showed that miR-34c-3p and miR-154-3p act as the most promising models in the diagnosis of metastatic osteosarcoma. Validation for this model by RT-qPCR in benign tissue and osteosarcoma biopsies confirmed the lower expression of miR-34c-3p and miR-154-3p in OS samples. In addition, a direct correlation between miR-34c-3p expression, miR-154-3p expression and tumor grade was discovered. The combined values of miR-34c-3p and miR-154-3p showed 90 % diagnostic power (AUC = 0.90) for osteosarcoma samples and 85 % (AUC = 0.85) for metastatic osteosarcoma. Adhesion junction and focal adhesion pathways, as well as epithelial-to-mesenchymal transition (EMT) GO terms, were identified as the most significant KEGG and GO terms for the top miRNAs. The findings of this study highlight the potential use of novel miRNA expression signatures for early detection of metastatic osteosarcoma. These findings may help in determining therapeutic approaches with a quantitative and faster method of metastasis detection and also be used in the development of targeted molecular therapy for this aggressive cancer. Further research is needed to confirm the clinical utility of miR-34c-3p and miR-154-3p as diagnostic biomarkers for metastatic osteosarcoma. • The impact of miRNA expression profiles on osteosarcoma's metastasis is explored. • Diverse datasets were analyzed using feature selection algorithms and machine learning. • Our study emphasizes the role of miRNAs in early diagnosis of metastatic osteosarcoma. • This research is a beginning for the future investigations into targeted molecular diagnostic. [ABSTRACT FROM AUTHOR]

This study presents an in-depth analysis of a global estimation model for compressive strength in concrete adopting various machine learning algorithms. Datasets encompass diverse features such as cement type, aggregate characteristics, and curing conditions, including extreme scenarios. Two feature selection methods were applied, and machine learning models such as Artificial neural networks (ANNs), Gaussian process regression (GPRs), support vector machines (SVMs), regression trees, and Ensembles were trained, validated, and tested. Findings indicate that model performance is influenced by both feature selection and dataset characteristics. ANNs and GPRs excelled across the evaluation metrics, with the top-performing models achieving R2 values greater than 0.9 from approximately 3000 datasets and maintaining deviations between predicted and actual data within a 20% margin. Partial dependency plots (PDPs) were also used to investigate relationships between features and compressive strength. A parametric study also qualitatively confirmed reasonable strength predictions were achieved within commonly used mixtures. The study recommends enhancing modeling precision through further inclusion of out-of-range data and expressing results probabilistically for future research. • Robust machine learning models predict concrete compressive strength. • Extensive dataset with diverse mix proportions and curing conditions. • Feature selection impact varies; independence among indicators noted. • Parametric study ensures reasonable strength predictions within common ranges. • Consideration of 'out-of-range' data enhances modeling precision and reliability. [ABSTRACT FROM AUTHOR]

• 18 fault types of AHU are assessed by impacts to symptoms. • 18 fault types are categorized into three levels based on symptom severity. • Various machine learning algorithms are used for FDD and performances were evaluated. • The new FDD framework improves diagnostic accuracy and efficiency. This study proposes a comprehensive framework for Fault Detection and Diagnosis (FDD) in Air Handling Units (AHU), emphasizing the impact of symptoms associated with various faults. The aim is to address an important limitation in FDD model development where detection is based simply on fault intensity without considering symptom impact or subjective severity criteria across faults. Instead, a new approach is introduced that factors in fault impact, utilizing impact analysis to enhance FDD accuracy and efficiency. The methodology involves three interconnected components: First, a fault impact analysis categorizes 18 fault types into stages per intensity and assesses resulting symptom impacts. Second, symptom-based intensity thresholds are established to categorize fault severity into three levels based on symptom severity. Finally, the goal is to integrate these findings into the FDD process for optimized fault detection and diagnosis. The study successfully categorized 18 faults into three severity levels, with thresholds identified for each. Furthermore, Tree-based models demonstrated effective performance when conducting FDD based on these levels. The approach provides a clear framework through three linked processes. This research combines comprehensive impact analysis with sophisticated classification methods, presenting a more defined, systematic, and efficient FDD approach. This novel methodology substantially advances fault detection and diagnosis, especially for building energy systems. [ABSTRACT FROM AUTHOR]

This research explores the use of machine learning (ML) models to forecast optical characteristics in photonic crystal fibers (PCF). Specifically, we focus on a solid core index-guided PCF with a hexagonal cladding arrangement. The primary challenges to PCF propagation analysis and predictions are accuracy, computational error, and time constraints. To address these difficulties, we have specially used ML ensemble models including Decision Tree Regressor (DTR), Random Forest Regressor (RFR), Gradient Boosting Regressor (GBR), eXtreme Gradient Boosting Regression (XGBR), and Bagging Regressor (BR). Model performance is assessed using metrics like Mean Squared Error (MSE) and R-squared (R2) through 10-fold cross-validation. Our key findings show that the GBR model outperforms other models and shows extremely low MSE and outstanding R2 values in predicting effective refractive index (Neff), effective mode area (Aeff), confinement loss, and dispersion. In addition, the study compares the performance of ML models with that of previous works using Artificial Neural Network (ANN), demonstrating improved efficiency in predicting optical characteristics for hexagonal PCFs. [ABSTRACT FROM AUTHOR]

Highway state agencies incur significant budget savings through optimal allocation of pavement Maintenance, Rehabilitation, and Reconstruction (MR&R) activities. These activities require robust prediction models that can handle large-scale, real-world data and can forecast pavement performance in the long run. Unfortunately, the traditional performance prediction models have been questionable in terms of efficiency and accuracy, are based on a limited number of explanatory variables, and are designated to predict short-term (up to five years) pavement conditions. Therefore, the goal of this study was to propose a machine learning-based technology that can predict the field performance by up to 11 years of Asphalt Concrete (AC) overlays placed on asphalt pavements in Southern states in the US based on key project conditions. The proposed technology resulted from assessing the prediction accuracy of machine learning algorithms, including Decision-Tree (DT), eXtreme Gradient Boosting (XGBoost), Artificial Neural Network (ANN), and ensemble-learning method, in forecasting the Pavement Condition Index (PCI) as the pavement performance indicator. For each algorithm, six models were developed sequentially based on historical pavement condition data collected from the Louisiana Department of Transportation and Development (LaDOTD) Pavement Management System (PMS) database. The six models learned from 892 log miles of randomly placed AC overlay sections in Louisiana. The output of these models was the future PCI of AC overlays at a biannual rate from one to 11 years. The findings showed that XGBoost and ensemble learning showed similar performance during model training and were further evaluated using the testing dataset. During model testing, the ensemble learning method yielded higher prediction accuracy than other algorithms, with R2 values decreasing from 0.77 at age 1 to 0.67 at age 11, Root Mean Square Error (RMSE) values increasing from ±1.65 to ±4.74, and Mean Absolute Error (MAE) increasing from 1.24 to 4.59. The resulting models can be used by state agencies to replace traditional prediction techniques. • The traditional performance prediction models have showed questionable efficiency and accuracy. • Longer prediction period of pavement performance index models can help in accurate pavement Maintenance, Rehabilitation, and Reconstruction (MR&R) budget allocation. • There is an urge need to develop a robust machine learning based technology to forecast the long-term field performance (up to 11 years) of AC overlays placed on asphalt pavements under hot and humid climate conditions, and successfully consider the impact of key pavement performance predictors. • The prediction accuracy of six machine learning algorithms, such as including Decision-Tree (DT), eXtreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN) was evaluated. [ABSTRACT FROM AUTHOR]

• Coupled high-res remote sensing with machine learning to assess agricultural pollution. • Establish monitoring network, optimize model parameters for improved simulation accuracy and reliability. • Used high-res data for detailed spatial and temporal analysis of pollution sources. • Identified key pollution sources and offered targeted mitigation strategies. • Proposed an adaptable monitoring framework for sustainable agricultural practices. Agricultural non-point source pollution threatens the quality of the ecological environment, human health, and safety. This study took the Sixth Drainage Ditch of the Yellow River Irrigation Area in Ningxia as the research area, set up a runoff water quality monitoring network, and comprehensively constructed an agricultural non-point source pollution monitoring model by combining the "source-sink" landscape theory, high-resolution remote sensing technology, and soil and water assessment tool (SWAT). The results showed that the simulation results of the flow and total nitrogen met the accuracy requirements. The R 2 values of total nitrogen in the calibration and validation periods were both > 0.8, and E ns was > 0.9. The regional applicability of the model was good. Based on the simulation results, the following conclusions were drawn. (1) The temporal distribution of the pollution load was concentrated in May–October, with peaks in June and August, which is consistent with the irrigation period. (2) Spatially, the pollution load was mainly distributed in sub-basins 1 and 5. The area is dominated by cultivated land and has poor conditions that are prone to nitrogen and phosphorus loss. (3) By quantitatively identifying pollution sources, the results showed that agricultural irrigation accounted for approximately 92.88 % of total pollutants. Compared with traditional methods, the monitoring method proposed in this study systematically evaluates the potential for non-point source pollution in the region and builds a relatively complete real-time monitoring network, improving data quality and model reliability. In addition, the relationship between river network density and catchment area threshold was used to optimize the catchment area threshold in the SWAT model, and non-point source pollution parameters suitable for the basin were obtained, providing a data basis and theoretical support for the large-scale application of the model. [ABSTRACT FROM AUTHOR]

One of the greatest challenges in mineral prospectivity mapping (MPM) research nowadays is to find a solid methodology that ensures the reliability of the prospectivity model during the learning and prediction procedures. Multiple uncertainties such as the location of non-deposit sites or the type of machine learning algorithm (MLA) can bias the MPM. To investigate these effects, we used multiple training datasets with different non-deposits locations, randomly created, and MLAs such as Artificial Neural Network (ANN), Random Forests (RF) and Support Vector Machine (SVM), to model orogenic-Au prospectivity in the Pitangui Greenstone Belt (PGB, Brazil). Regarding the implications in the methodology for MPM, there are great differences between the models' performances in mapping prospective zones when there is a slightly change in the location of negative samples. These changes can be observed by using the Shapley additive explanation metrics (SHAP values), which can help mitigate such effects by choosing an optimal model among all randomly created datasets. The SHAP values of non-deposit sites also showed that ANN and SVM present overfitting problems despite the use of balanced data. RF on the other hand outperformed in all ten datasets and showed great recognition and adjustment to the negative samples. The results presented in this research are also promising to the prospective studies in the PGB, as it shows a map capable to correctly predict 97 % of the known deposits and occurrences in 3 % of the total area and points the new frontiers for gold exploration in the PGB. • The Pitangui Greenstone Belt has great potential to host gold mineralization. • Location of negative samples influence the performance of prospectivity models. • Random Forest prospective models outstand in predict mineral deposits. • Structural framework and lithologies are the main proxies for gold exploration. [ABSTRACT FROM AUTHOR]

The efficient and accurate analysis of illicit drugs remains a constant challenge in Australia given the high volume of drugs trafficked into and around the country. Portable drug testing technologies facilitate the decentralisation of the forensic laboratory and enable analytical data to be acted upon more efficiently. Near-infrared (NIR) spectroscopy combined with chemometric modelling (machine learning algorithms) has been highlighted as a portable drug testing technology that is rapid and accurate. However, its effectiveness depends upon a database of chemically relevant specimens that are representative of the market. There are chemical differences between drugs in different countries that need to be incorporated into the database to ensure accurate chemometric model prediction. This study aimed to optimise and assess the implementation of NIR spectroscopy combined with machine learning models to rapidly identify and quantify illicit drugs within an Australian context. The MicroNIR (Viavi Solutions Inc.) was used to scan 608 illicit drug specimens seized by the Australian Federal Police comprising of mainly crystalline methamphetamine hydrochloride (HCl), cocaine HCl, and heroin HCl. A number of other traditional drugs, new psychoactive substances and adulterants were also scanned to assess selectivity. The 3673 NIR scans were compared to the identity and quantification values obtained from a reference laboratory in order to assess the proficiency of the chemometric models. The identification of crystalline methamphetamine HCl, cocaine HCl, and heroin HCl specimens was highly accurate, with accuracy rates of 98.4 %, 97.5 %, and 99.2 %, respectively. The sensitivity of these three drugs was more varied with heroin HCl identification being the least sensitive (methamphetamine = 96.6 %, cocaine = 93.5 % and heroin = 91.3 %). For these three drugs, the NIR technology provided accurate quantification, with 99 % of values falling within the relative uncertainty of ±15 %. The MicroNIR with NIRLAB infrastructure has demonstrated to provide accurate results in real-time with clear operational applications. There is potential to improve informed decision-making, safety, efficiency and effectiveness of frontline and proactive policing within Australia. [ABSTRACT FROM AUTHOR]

In this work, Surface-enhanced Raman spectroscopy (SERS) along with machine learning algorithms (MLA) were used to detect and classify the viral particles to assess the possibility of using the spectra of inactivated influenza A viruses for MLA training and spectra database compilation for further study and diagnosis of intact forms of the virus. Viral particles inactivation was performed by formalin, ultraviolet and beta-propiolactone. Support vector method and principal component analysis allowed to classify intact and inactivated viral particles spectra with an accuracy of 80.0–96.7 %. The results obtained suggest that it is not advisable to create a spectral database and train machine learning algorithms for their further application in SERS diagnostics of intact viruses based on the spectra of the inactivated virus particles. • SERS and ML algorithms detect and classify viral particles. • Investigated inactivation of influenza A viruses for MLA training. • Achieved 80.0-96.7% accuracy using SVM and PCA. • Caution against using spectra of inactivated viruses for SERS diagnostics. [ABSTRACT FROM AUTHOR]

• Exploring the impact of winter and summer LST changes on GPP from 2001 to 2020 on a global scale. • Linear regression model, ANN, Random forest, XGBoost algorithms, and Polynomial regression are used to discuss the relationship between LST and GPP. • The Sen+MK trend analysis model is used to analyze the changes in LST and GPP. Understanding the impact of global land surface temperature (LST) changes on gross primary production (GPP) is crucial for addressing global sustainability challenges effectively. This article explores the effects of winter and summer temperature variations on GPP from 2001 to 2020 on a global scale, employing a range of modeling techniques to investigate the complex relationship between temperature changes and GPP. The study employs various modeling approaches, including linear regression models, artificial neural networks, Random Forest (RF) models, and the XGBoost algorithm to examine both linear and non-linear relationships between global temperature changes and GPP. The results reveal that the RF and XGBoost models effectively capture the non-linear relationship between LST and GPP during both summer and winter, demonstrating a high level of statistical significance (P < 0.01) and achieving R2 values of 0.598. Furthermore, the study applies a Sen+MK trend analysis model to identify six distinct trend patterns in LST and GPP during both summer and winter seasons. In summer, the area exhibiting a non-significant increasing trend (NSI) for GPP covers 65,066,274.77 km2, whereas in winter, a strong significant decreasing trend (SSD) for GPP spans 64,537,108.31 km2. Notably, GPP patterns closely mirror those of LST, suggesting that rising high-temperature conditions during summer lead to reduced GPP, while increased low temperatures during winter promote GPP growth. Robust correlations between LST and GPP are observed under various trend conditions in both summer and winter, displaying strong statistical significance (P < 0.01), with SSD achieving a maximum R2 of 0.594. These findings contribute significantly to our comprehensive understanding of the dynamic relationship between LST and GPP and offer valuable insights for addressing the sustainable development of global climate change challenges. [ABSTRACT FROM AUTHOR]

The construction industry's growing emphasis on sustainability has driven the development of eco-friendly concrete alternatives, such as self-compacting concrete (SCC) and geopolymer concrete (GPC). These materials, incorporating diverse ingredients like fly ash, silica fumes, and recycled aggregates, aim to reduce the environmental impact of traditional concrete production. However, the complex behavior of these materials poses challenges in accurately predicting their properties. Machine learning (ML) and deep learning (DL) techniques offer promising solutions to this problem. This systematic literature review analyzes the application of ML and DL methods for predicting the properties of sustainable concrete, including SCC, GPC, and other variations incorporating recycled aggregates, supplementary cementitious materials, and high-performance concrete. The review encompasses linear models, support vector machines (SVM), k-nearest neighbors (KNN), bagging, boosting, artificial neural networks (ANN), evolutionary algorithms, and nature-inspired algorithms. The analysis reveals a dominance of ANNs in predicting GPC properties, while bagging, boosting, and ANNs perform well for SCC (60 %). For other sustainable concrete types, boosting methods are prominent. The review highlights the potential of AI to optimize concrete mix designs, enhance sustainability, and reduce reliance on costly and time-consuming experimental testing. Additionally, it identifies key challenges and knowledge gaps, paving the way for future research in this rapidly evolving field. • AI models, particularly ANNs, excel at predicting mechanical properties across various concrete types. • Hybrid models combining ANNs with nature-inspired algorithms improve prediction accuracy. • Durability prediction, especially for SCC and GPC, receives less attention and requires further research. • Current datasets, particularly for GPC, might be too small, limiting model accuracy and generalizability. • Use AI to predict durability (particularly fire resistance) and for ultra-high-performance, self-healing, and nano concrete. [ABSTRACT FROM AUTHOR]

The eutrophication pollution caused by excessive phosphorus seriously affects production and daily life of the people. Accurate and rapid detection of phosphate is crucial for water quality management and cleaner production. Currently, most analytical methods for phosphates involve the addition of multiple chemical reagents and are incapable of simultaneous quantification of total phosphorus (TP) and phosphate anions. The environmental friendliness and analytical performance of the methods need to be improved. Here, we established a green and high-accuracy quantitative model based on absorption spectroscopy and machine learning algorithms, which achieved the analysis of TP, H 2 PO 4 − and HPO 4 2−. Based on the pH and absorption spectra of phosphate anions, classification and regression tree (CART), random forest (RF), and extreme gradient boosting (XGBoost) algorithms were used to train the original absorption spectra data, first-order derivative spectra data, and multiple feature parameter data of phosphate solutions with pH 4.7–10.0. By comparing the R2, root mean square error (RMSE) and mean absolute error (MAE) evaluation indicators, it was found that the phosphate anions quantitative model established using the CART based on the feature parameters of 187 nm performed the best, R2 > 0.9999, RMSE <3.4 × 10−6, MAE <8.1 × 10−7. The quantitative ranges of TP, H 2 PO 4 −, and HPO 4 2− were 3.98 × 10−4—9.33 × 10−4 mol/L, 5.98 × 10−6—4.20 × 10−4 mol/L and 1.53 × 10−6—7.43 × 10−4 mol/L, respectively. Moreover, the quantitative model was optimized based on the spectral characteristics of 14 interfering ions, to improve the model analytical performance for unknown samples. The quantitative model developed in the study achieves accurate analysis of TP, H 2 PO 4 − and HPO 4 2−, which is in line with green chemistry concept, and has the potential for prediction and early warning of water environment quality. [Display omitted] [ABSTRACT FROM AUTHOR]

• The demosaiced RGB imagery can improve the estimation of E. coli spatial patterns. • The quantile-based data splitting improved modeling of Imbalanced E. coli datasets. • The visible blue intensity and air temperature were the most influential inputs. The rapid and efficient quantification of Escherichia coli concentrations is crucial for monitoring water quality. Remote sensing techniques and machine learning algorithms have been used to detect E. coli in water and estimate its concentrations. The application of these approaches, however, is challenged by limited sample availability and unbalanced water quality datasets. In this study, we estimated the E. coli concentration in an irrigation pond in Maryland, USA, during the summer season using demosaiced natural color (red, green, and blue: RGB) imagery in the visible and infrared spectral ranges, and a set of 14 water quality parameters. We did this by deploying four machine learning models – Random Forest (RF), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGB), and K-nearest Neighbor (KNN) – under three data utilization scenarios: water quality parameters only, combined water quality and small unmanned aircraft system (sUAS)-based RGB data, and RGB data only. To select the training and test datasets, we applied two data-splitting methods: ordinary and quantile data splitting. These methods provided a constant splitting ratio in each decile of the E. coli concentration distribution. Quantile data splitting resulted in better model performance metrics and smaller differences between the metrics for both the training and testing datasets. When trained with quantile data splitting after hyperparameter optimization, models RF, GBM, and XGB had R 2 values above 0.847 for the training dataset and above 0.689 for the test dataset. The combination of water quality and RGB imagery data resulted in a higher R 2 value (>0.896) for the test dataset. Shapley additive explanations (SHAP) of the relative importance of variables revealed that the visible blue spectrum intensity and water temperature were the most influential parameters in the RF model. Demosaiced RGB imagery served as a useful predictor of E. coli concentration in the studied irrigation pond [Display omitted] [ABSTRACT FROM AUTHOR]

• Accurate machine learning models minimize delays and losses in manufacturing. • Cryogenically treated electrodes boost EDM wear prediction accuracy. • Identified influential factors that optimize EDM processes. • Enhancing productivity and quality in manufacturing. • Future of manufacturing: advanced techniques for growth. In manufacturing, predicting and reducing electro-erosion wear during the electric discharge machining (EDM) process is critical to minimize delays, financial losses and product defects. Achieving this requires developing and evaluating accurate machine learning models. In our study, we focus on cryogenically treated mold steel electrodes to investigate the potential of different machine learning algorithms to predict EDM wear. We considered five machine learning algorithms—artificial neural networks, ensemble learning, boosting algorithms, tree-based algorithms, and k-nearest neighbors—to evaluate their ability to predict wear patterns accurately. Each algorithm was trained and tested using actual experimental data from EDM processes. Our results show that the machine learning models demonstrated exceptional accuracy, accurately predicting EDM wear in training and testing datasets with almost 99% accuracy. In addition, we identified the most influential characteristics that affect wear patterns, including operating current, cryogenic process parameters, and electrode composition. Based on these findings, manufacturers can gain valuable insight into the factors that cause EDM wear and optimize their EDM processes accordingly to improve productivity, reduce wear-related costs, and increase production quality across multiple manufacturing industries. Furthermore, this research provides insights into the possibilities of implementing these models in real manufacturing contexts and motivates future research on this topic. Ultimately, integrating advanced computing techniques and prudent decision-making strategies will shape the future of manufacturing operations management and promote sustainable and profitable business growth. [ABSTRACT FROM AUTHOR]

[Display omitted] • The high consumption of maize requires primarily the identification of transgenic products. • Elemental Spectroscopy techniques provide sufficient information about the samples. • Both non-supervised and supervised multivariate analyses are effective for discrimination. • Internal and external validation provide reliability for classification tasks. In the last decades, the production and consumption of genetically modified agricultural products have increased markedly due to the worldwide population growth and rising demand for food and feed. Consequently, genetically modified crops have been extensively produced and consumed, which required identifying and discriminating transgenic and non-transgenic products. Laser-induced breakdown spectroscopy (LIBS) combined with chemometric methods was applied to identify and discriminate two varieties of conventional (not-genetically modified, NGM) maize from four varieties of transgenic maize (genetically modified, GM). The LIBS spectra acquired under reduced pressure (100 Torr) conditions over two ranges, i.e., 175–330 nm and 275–770 nm, were subjected to Standard Normal Variation (SNV) and multivariate methods such as Principal Component Analysis (PCA) to reduce data matrices dimensionality and spectral noise. The supervised machine learning algorithms k-nearest neighbor (k-NN) and support vector machine (SVM) have been applied to discriminate among NGM and GM maize reserving 25 % of data for external validation. The training data were employed for hyperparameter optimization of classifiers using the Leave-One-Out Cross-Validation (LOOCV) method. Considering all six maize varieties simultaneously, the highest training accuracy achieved was 90.56 %, with an external validation accuracy of 88.33 %. In an alternative approach based on pairwise combinations of one GM variety against one NGM variety, the best outcome achieved was 100 % LOOCV and external validation accuracy. These results showed that LIBS supported by appropriate chemometric methods represents an alternative screening technique for identifying and discriminating transgenic from non-transgenic maize. [ABSTRACT FROM AUTHOR]

This paper aims to explore the utilization of machine learning techniques for the accurate prediction of rheological properties in a specific nanofluid system, ZnO(50 %)-MWCNTs (50 %)/Ethylene glycol (20 %)-water (80 %), designed for nano-refrigeration applications. The effective manipulation of the rheological behavior of nanofluids is pivotal for enhancing their heat transfer efficiency and overall performance. By harnessing the predictive power of machine learning, this study endeavors to unravel the intricate relationships governing the rheological characteristics of the nano-refrigerant, ultimately contributing to the development of advanced cooling solutions. The obtained results show that μ nf of ZnO(50%)-MWCNTs (50%)/ Ethylene glycol(20%)-water(80%) nano-refrigerant is little affected by T, and even when T varies, this result does not alter much. Also, the lowest μ nf occurs when it has the highest temperature and the lowest γ and φ. Finally, it was concluded that the best algorithm in terms of the Taylor diagram for μ nf output is the MPR algorithm and the worst is the ECR algorithm and the pattern of γ changes shows that the ideal value of γ is the biggest when μ nf levels fall in tandem with their growth. [ABSTRACT FROM AUTHOR]

In this study, we apply and compare different algorithms from machine learning to describe and predict NEET rates in 31 European countries in the period from 2005 to 2020. With this aim, we considered eleven indicators describing the socio-economic national context and the level of innovation of the economies. Besides improving knowledge about the use of machine learning algorithms for the description of the NEET phenomenon, we discuss the connections between NEETs and other indicators that connect with other relevant sustainable development goals (SDGs), such as education, the reduction of inequalities, and decent work for everyone. The reduction of NEET rates is the only goal directly addressed to young people, The article underscores the need for evidence-based approaches to measure SDG achievement, especially concerning the heterogeneous NEET population. It emphasizes the importance of machine learning algorithms as a modern methodology for understanding and addressing the NEET phenomenon within the framework of SDGs, considering the complex interrelationships of socio-economic factors contributing to social and economic sustainability. • We identify the Best Machine Learning Algorithms for NEET and SDG studies. • We connect NEET paths in 2005–2020 with other relevant SDG goals. • We propose a complex approach, combining 31 countries and various indicators. [ABSTRACT FROM AUTHOR]

• Determination of the necessary features set from the raw EEG signals. • Use of Energy, Power and Entropy based feature extraction techniques. • Classification of the features set using LDA, QDA, Naïve Bayes, and Decision Tree. • Selection of the best methodology for classification of the MI-based EEG signals. The initial principal task of a Brain-Computer Interfacing (BCI) research is to extract the best feature set from a raw EEG (Electroencephalogram) signal so that it can be used for the classification of two or multiple different events. The main goal of the paper is to develop a comparative analysis among different feature extraction techniques and classification algorithms. In this present investigation, four different methodologies have been adopted to classify the recorded MI (motor imagery) EEG signal, and their comparative study has been reported. Haar Wavelet Energy (HWE), Band Power, Cross-correlation, and Spectral Entropy (SE) based Cross-correlation feature extraction techniques have been considered to obtain the necessary features set from the raw EEG signals. Four different machine learning algorithms, viz. LDA (Linear Discriminant Analysis), QDA (Quadratic Discriminant Analysis), Naïve Bayes, and Decision Tree, have been used to classify the features. The best average classification accuracies are 92.50%, 93.12%, 72.26%, and 98.71% using the four methods. Further, these results have been compared with some recent existing methods. The comparative results indicate a significant accuracy level performance improvement of the proposed methods with respect to the existing one. Hence, this presented work can guide to select the best feature extraction method and the classifier algorithm for MI-based EEG signals. [ABSTRACT FROM AUTHOR]

Psychological responding to stressful situations such as experiential avoidance (EA) is central to the development of several psychopathologic conditions, yet it tends to be assessed solely via retrospective self-report. The purpose of this study is to utilize machine-learning algorithms to delineate and predict individuals classified as high vs. low in EA across stress-inducing situations based on their psychophysiological signals. Three samples (total N = 212) arising from previous research included 43 smokers, 86 females at high-risk for eating disorders, and 83 anxious individuals. Each of the studies involved a stress induction laboratory experiment. Supervised algorithms were used to compare the ability of physiological signals (skin conductance (SCR), heart rate variability (HRV), and facial electromyography (fEMG)) in classifying both individuals presenting with high vs. low EA (using common self-report measures) and situations. Overall accuracy of the Rectangular Window Methodology (RWM) model was 77.8%, with prediction of 88.9% accuracy of participants being high in EA. The features of SCR, fEMG and HRV accurately and stably classified individuals characterized by dispositional high EA as they navigated stressful situations. This study is the first to use machine learning to delineate and predict high vs. low EA in individuals and situations from psychophysiological signals. These parameters can be used for future real-time assessments and predictions. • Study utilized existing data to analyze via machine learning algorithms experiential avoidance. • Machine learning algorithms had 88.9% prediction accuracy of high experiential avoiders. • Machine learning algorithms can be used to predict experiential avoidance. [ABSTRACT FROM AUTHOR]

The progression of the fuzzy domain is propelled synergistically by the evolution of science via academic papers and the advancement of technology through technical patents. However, the intrinsic interrelationships within the science and technology (S&T) interactive landscape of this domain remain largely unknown, as few studies attempt to explore these close interrelationships. To bridge this research gap, based on 161,060 academic publications and 36,693 granted patents, we measured their mutual attribution between S&T from the perspective of textual semantics and structure information on top of a quantitative network monitoring process. The results showed that the fuzzy domain lacked technology-oriented topics in terms of semantic relations and presented an interactive pattern where science was ahead of technology; that is, science worked as an engineer to expand the knowledge boundaries of the research field, and technology existed more as a beneficiary. This article presents cutting-edge insights into the intrinsic S&T interactions of the fuzzy domain in a visual and quantitative approach for the first time and aims to aid researchers in theoretically grasping the S&T innovation laws. [ABSTRACT FROM AUTHOR]

• Experimentation of AWJ milling of alumina ceramic using Box-Behnken RSM. • Data augmentation using a bootstrapping technique. • Hyperparameter tuning using a grid search with 5-fold cross validation. • Development of machine learning algorithms using the augmented data samples. • The methodology was tested on dataset of WEDM of metal matrix composites. In abrasive waterjet (AWJ) milling process, the ability to accurately predict outcomes such as surface roughness (R a) and depth of cut (DoC) holds immense significance for cost reduction, resource optimization, and minimizing material wastage. However, a major challenge lies in the limited availability of AWJ milling datasets, hindering the development of robust machine learning models. To address the predicament of outcome, the present study employs a bootstrapping data augmentation technique to expand small AWJ milling datasets. Augmented data is used to train machine learning models, and hyperparameter fine-tuning is conducted using grid search and 5-fold cross-validation. In the prediction of surface roughness (R a) for AWJ milling of alumina ceramic, the study finds that a Multi-Layer Perceptron (MLP) Regression model with a 4–100-100–1 architecture, 'relu' activation function, 'adam' solver, alpha value of 0.0001, and 'adaptive' learning rate outperforms the decision tree and random forest regression models. In the case of prediction of AWJ milling depth of cut (DoC) in alumina ceramic, the random forest regression model surpasses MLP and decision tree regression models. It employs hyperparameters including 100 estimators, 'gini' criterion, 'log2′ for maximum features, a maximum depth of 50, a minimum sample split of 10, a minimum sample leaf of 2, bootstrap as 'True', and a minimum impurity decrease of 0. The methodology followed to develop the machine learning models in this study is also tested in wire electrical discharge machining (WEDM) process in the machining aspects of metal matrix composites (MMC) for the prediction of surface roughness (R a). The experimental dataset was adopted from (Satishkumar et al., 2011). In case of prediction of surface roughness (R a) in WEDM, the random forest regression model outperforms MLP and decision tree regression models. These developed machine learning models hold the potential to greatly assist manufacturing engineers in selecting optimal input parameters for AWJ milling and WEDM, thereby reducing the wastage of machine resources. [ABSTRACT FROM AUTHOR]

• Multi-sensor input machine learning vastly outperforms single-senor input models. • Recursive feature elimination reduces machine learning model complexity. • Gaussian process regression model best estimates wheat nitrogen nutrition index. • Transfer component analysis was used to improve model transferability. The nitrogen nutrition index (NNI) has been extensively applied for the diagnosis of crop nitrogen status, providing insights into efficient nitrogen utilization and plant growth. In this study, we utilized a low-altitude unmanned aerial vehicle (UAV) platform, equipped with multispectral (MS), red–green–blue (RGB), and thermal infrared (TIR) cameras, to comprehensively capture wheat spectral information. The analysis of the relationship between NNI and relative yield revealed an initially linear relationship, which saturated for high NNI values. To enhance accuracy and minimize complexity, we employed a random forest (RF) – recursive feature elimination (RFE) method to select features as inputs for four machine learning (ML) models: back propagation neural network (BPNN), extreme learning machine (ELM), support vector regression (SVR), and Gaussian process regression (GPR). After feature selection, the prediction accuracies of single-sensor models were ranked as: MS > RGB > TIR. The R2 values for the four ML models were in the range of 0.54–0.75. Among multi-sensor combinations, the GPR with MS + RGB + TIR input features achieved the best results with R2 = 0.89 and RPD = 2.52. Further, the dataset was partitioned into six subsets based on location and cultivar variety to evaluate model transferability. The results showed that the transferability largely suffered during the bivariate conditions of different varieties at different locations; the transferability of the model was average improved by 11 % when GPR was combined with transfer component analysis (TCA). The accuracy and transferability of the NNI estimation models significantly improved, offering valuable guidance and methodological support for diagnosing the nitrogen nutrient status of wheat. [ABSTRACT FROM AUTHOR]

• DSPA and CNN are the best ML algorithms for sepsis prediction. • Imbalance-XGBoost outperforms other traditional ML algorithms. • DSPA can improve sepsis diagnosis and management. • Further research needed on poor-performing ML algorithms. With the integration of artificial intelligence and clinical medicine, machine learning (ML) algorithms have been applied to develop sepsis predictive models for sepsis management. The purpose is to systematically summarize existing evidence to determine the effectiveness of ML algorithms in sepsis. We conducted a systematic electronic search of databases including PubMed, Cochrane Library, Embase, and the Web of Science, and included all case-control and cohort studies using terms reflecting sepsis and ML up to September 2023. statistical software STATA was used for network meta -analysis, and QUADAS-2 tool was used to assess the certainty of evidence. The SUCRA results for sensitivity, specificity, and predictive accuracy of various models are as follows: DSPA (77.0 %) > Imbalance-XGBoost (72.9 %) > CNN + Bi-LSTM (69.7 %) > CNN (67.3 %) > LR (62.4 %) > Ensemble model (55.9 %) > RF (53.2 %) > ET (51.3 %) > XGBoost (49.1 %) > DNN (48.1 %) > MLP (47.5 %) > RBF (47.1 %) > KNN (45.8 %) > NB (33.3 %) > SVM (13.7 %) > Bi-LSTM (5.7 %); CNN (78.3 %) > CNN + Bi-LSTM (77.6 %) > DSPA (75.1 %) > ET (69 %) > Bi-LSTM (68.5 %) > MLP (51 %) > RBF (50.2 %) > KNN (47.3 %) > RF (47 %) > Ensemble Model (43.4 %) > XGBoost (38.1 %) > SVM (37.3 %) > NB (34.2 %) > DNN (31.1 %) > LR (30.4 %) > Imbalance-XGBoost (21.5 %); DSPA (85.9 %) > CNN + Bi-LSTM (82.6 %) > CNN (81.9 %) > Imbalance-XGBoost (76.8 %) > ET (67.8 %) > RF (51.1 %) > Ensemble model (47.7 %) > XGBoost (44.4 %) > LR (42.7 %) > MLP (38.1 %) > RBF (37.8 %) > KNN (37.3 %) > DNN(35.8 %) > Bi-LSTM(33.3 %) > NB(21.5 %) > SVM(15.3 %). DSPA and CNN may be the best ML algorithms for predicting sepsis. Imbalance-XGBoost algorithm outperformed other traditional ML algorithms in terms of sensitivity and predictive accuracy. This study has several implications for clinical practice and research, highlighting the potential benefits of using ML algorithms in sepsis management, particularly in improving sepsis detection and reducing mortality rates. Through our systematic review and network meta -analysis, we have provided a comprehensive and accurate assessment of the effectiveness of ML algorithms in sepsis prediction, emphasizing the need for further exploration and evaluation of these algorithms to advance sepsis management. [ABSTRACT FROM AUTHOR]

• Simulate and analyze the characteristics of LULC changes from 2000 to 2030. • UTFVI is used to describe the thermal environment of the PRD urban agglomeration. • Machine learning algorithms are applied to predict the seasonal LST, UTFVI, and UHI. • Explore the response between urban land expansion and the thermal environment. Land use practices in urban areas exert a profound influence on the urban thermal environment and the pursuit of sustainable development. This paper aims to investigate and forecast future changes in land use/land cover (LULC) and their response to seasonal variations in land surface temperatures (LST), the urban thermal field variance index (UTFVI), and the urban heat island effect (UHI). The artificial neural network based on cellular automata (ANN-CA) and the improved whale optimization based on long short-term memory (WOA-LSTM) algorithms are used to predict the LULC, UTFVI, and UHI characteristics in the Pearl River Delta (PRD) urban agglomeration. The results show that urban land will likely expand from 4335 km2 to 8292 km2 from 2000 to 2030. The LST continues to increase, and the maximum temperature in summer will likely increase to 44.6 °C in 2030. Without the intervention of effective cooling measures, the area with LST≥35 °C will likely increase to 4873 km2, and the proportion of areas with LST≥20 °C will likely reach 63.72 % in the winter of 2030. The strongest level of UTFVI expansion is significant in summer, and the area is likely to increase by 83.64 % in 2030. Urban land has the highest percentage in the high temperature region relative to other land use categories. Similar to the trend of LST changes, UHI is expected to notably increase by 2030, with minimum and maximum UHI values projected to rise. This study may provide new perspectives on thermal environment management and sustainable urban development. [ABSTRACT FROM AUTHOR]

[Display omitted] • An automatic sorting machine is designed for grading pomegranates. • Machine vision is utilized in order to detect sunburn on pomegranates. • Three different image processing algorithms are proposed for feature extraction. • ANN and SVM classification models are used to sort the fruits automatically. • The experimental results of algorithms and overall sorting machine are provided. Sunburn is a widespread problem in pomegranate marketing. About 30 % of the harvested pomegranates suffer from sunburn. In this research, an automatic sorting machine was designed, developed, and tested to separate the sunburn pomegranates. To detect sunburn on pomegranates, three novel image processing algorithms were proposed: skin spot detection, brightness intensity distribution, and statistical surface parameter. Two supervised machine learning algorithms comprising artificial neural networks and support vector machines were utilized to classify the fruits. The proposed algorithms were used to control the developed sorting machine to separate the sunburn pomegranates from the healthy ones. Finally, the performance of the proposed algorithms was evaluated by the accuracy and response time of the sunburn sorting function of the developed machine. The brightness intensity distribution algorithm with the accuracy and response time of 98 % and 0.88 s, respectively, is suggested as the fastest and the most accurate algorithm. [ABSTRACT FROM AUTHOR]

[Display omitted] • The impact of the DTPA chelating agent on the rock/oil contact angle and rock surface charge was examined under varying concentrations, salinity, and potential-determining ions. • An experimental database comprising 240 sets of wettability data was acquired. • Two advanced machine-learning models, namely Random Forest and Boosted Regression Tree, were developed to accurately estimate the alteration in rock wettability. • Statistical analysis was conducted, demonstrating outstanding predictions for a wide range of input variables. Chelating agents have garnered increasing attention due to their unique capabilities in addressing challenges associated with chemical enhanced oil recovery (CEOR) materials. This research investigates the transformative potential of diethylenetriaminepentaacetic acid (DTPA) chelating agent in altering sandstone rock wettability. Wettability and zeta potential measurements were used to assess the impact of crucial factors including DTPA concentration, salinity, and potential determining ions (PDIs) on contact angle and rock surface charges. The findings revealed that there is an optimal level for DTPA concentration and salinity that can shift rock wettability from an oil-wet to strongly water-wet state. Moreover, while PDIs in their natural concentrations had no significant impact on DTPA performance, a threefold increase in their concentration was found to adversely affect DTPA efficacy. After conducting rock wettability tests, two advanced machine learning (ML) techniques, namely Random Forest (RF) and Boosted Regression Tree (BRT), were employed to assess rock/oil contact angle based on the aforementioned factors. Data were collected from 240 experimental datasets of contact angles. The findings indicated that both models performed well, but BRT exhibited exceptional performance. Sensitivity analysis, conducted using the Jackknife method, revealed the order of importance for parameters affecting wettability as follows: PDIs > salinity > time > DTPA concentration. [ABSTRACT FROM AUTHOR]

[Display omitted] • The impact of diethyl ether on the interfacial tension (IFT) of crude oil-brine was examined under varying salinity and temperature conditions. • An experimental database comprising 7,017 sets of crude oil/brine/DEE IFT data was acquired. • Six advanced machine-learning models were developed to accurately estimate the IFT of Crude oil–Brine-DEE. • Statistical analysis was conducted, demonstrating outstanding predictions for a wide range of input variables. Solvent injection, a well-established method for enhanced oil recovery (EOR), has demonstrated significant improvements in oil recovery when compared with conventional water flooding techniques. The interfacial tension (IFT) is pivotal in determining the displacement efficiency and overall performance of innovative techniques like dimethyl ether-enhanced waterflooding (DEW), which has gained substantial attention in recent years. In this study, following laboratory measurements of IFT, six advanced machine learning (ML) techniques were employed: Generalized Linear Model (GLM), Generalized Additive Model (GAM), Random Forest (RF), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), and Boosted Regression Tree (BRT) to model the IFT in both oil-brine and oil-brine-diethyl ether (DEE) systems. The analysis is based on an extensive dataset comprising 7,017 data points for oil-brine and 6,949 data points for oil-brine-DEE systems obtained from experimental studies. The findings indicate that the developed RF model excels in predicting IFT, boasting a remarkable coefficient of determination (R2 = 0.99) along with the lowest root mean squared error (RMSE = 0.2), mean squared error (MSE = 0.04), and mean absolute error (MAE = 0.13). The study underscores the significance of optimizing salinity levels to achieve the most substantial reduction in IFT. This reduction is attributed to the enhanced migration of polar components, such as asphaltene molecules, to the interface of the oil-brine system. Moreover, the research highlights a synergistic decrease in IFT when both DEE and soluble ions are present, resulting in the lowest IFT at around 2 mN/m in 40,000 ppm salinity (S 2) at 70 °C (T 3). This indicates that the adsorption of DEE at the water–oil interface forms a layer capable of adsorbing ions, thereby enhancing the layer's thickness. As a result, the oil–solvent-ion layer becomes thicker compared to the oil-ion layer, leading to the maximum decrease in IFT. Additionally, with increasing temperature up to 70 °C, the IFT of both systems demonstrated a downward trend, as evidenced by all experiments. The outcomes of this study have the potential to enhance our comprehension of the underlying mechanisms involved in water-soluble solvent EOR techniques. [ABSTRACT FROM AUTHOR]

• This study uses descriptive statistical features and five classifiers—DT, LDA, NB, SVM, and KNN—to recognise hand gestures. • Five classifiers were tested in two situations by dividing the dataset into 80%-20% and 50%-50% training and testing ratios. • The SVM classifies 8 hand gestures with the maximum accuracy at %97.1 for 80%-20% and %96.1 for 50%-50%. • Using seven performance metrics, the five classifiers showed strength and precision with good numerical values. • Our research outperformed nine previous studies using the same MMG signal, demonstrating superior classification accuracy. One of the most significant tools for Human–Computer Interaction (HCI) is body language, namely gestures. Therefore, Gesture recognition using Mechanomyography (MMG) for controlling prostheses and rehabilitation devices has been researched for years. The MMG is gauged as the mechanical vibrations engendered by the muscle in its activation. Hence, MMG data provides information regarding the mechanical activity of muscles. It is obtained from the limb and facilitates the comprehension of hand gestures. A MMG dataset named "MMG DATA" that has already been produced is utilized in this paper. To improve the accuracy of classification and the speed of processing, we have extracted the twelve descriptive statistical features of the eight game motions and their iterations contained in the "MMG DATA" for each subject. Five distinct machine learning algorithms—including Decision Tree (DT), Linear Discriminant Analysis (LDA), Naive Bayes (NB), Support Vector Machine (SVM), and k-Nearest Neighbor (k-NN)—were used to categorize the hand gestures, with the results compared using several performance metrics. Several experiments have been carried out in order to validate the presented models. Additionally, two scenarios have been achieved by splitting the dataset into two different and far apart ratios: the first by randomly partitioning the dataset into 80% for training the models and 20% for testing them, while the second by evenly dividing the dataset into 50% for both training and testing phases. Drawing on the results of the performance metrics, it can be argued that this study effectively distinguishes and classifies the eight hand motions. Furthermore, it is worth noting that the results exhibit significant strength and stability, particularly when transitioning from the initial scenario to the subsequent scenario across all proposed models. Out of the classifiers that were chosen, the SVM emerged as the most successful, achieving a classification accuracy of 97.1%. [ABSTRACT FROM AUTHOR]

It is nowadays a hot topic to apply machine learning (ML) algorithms to illustrate water temperature dynamics in lentic waters. Due to the limited amount of in-situ temperature measurements from traditional sampling programmes, most of the related studies, however, restricted their analysis within the surface and rarely checked the results for whole depth profiles. Moreover, capability of such methods in projecting thermal dynamics under future climate conditions is even less illustrated. To fill the gap, we collected an unparalleled huge database including 9 million water temperature measurements, from 2013 to 2022, in Rappbode Reservoir, Germany as well as the corresponding climatic and hydrological observations, to train three commonly used ML models (Random Forest, XGBoost, and Long Short-Term Memory) and comprehensively evaluate their performance in reproducing thremal structure within the whole water column. We also systemically compared such simulation results with a well-established process-driven model (CE-QUAL-W2). Going beyond a pure reproduction of observatiosn, we further evaluated the ability of CE-QUAL-W2 and ML models in projecting thermal dynamics under RCP8.5 climate scenario up to 2100. Our results suggested three ML methods yielded high accuracy in capturing water temperature dynamics from the surface (epilimnion) to bottom (hypolimnion), and also satisfactorily reproduced temporal development of the stratification pattern, with the results corresponding well with those from CE-QUAL-W2. By contrast, the projections by ML models were rather insensitive to future climate warming under RCP8.5 comparing with those by CE-QUAL-W2, pointing to an important risk whenever ML-based models are extrapolated beyond their training data range. Supported by millions of in situ measurements, our research clearly illustrated the opportunities (limits) of ML models in projecting thermal dynamics in deep waters, and the conclusion also provides key guidance for scientists to select the appropriate tools in projecting water temperature, and other environmental factors, under changing climate. [Display omitted] • An unparalleled temperature database was collected to test performance of ML models. • ML models captured the seasonal temperature pattern from top to bottom water layers. • ML models accurately reproduced the temporal development of stratification structure. • Extrapolation in ML models should be cautious under climate warming conditions. [ABSTRACT FROM AUTHOR]

Landslides pose a significant threat to lives and socio-economic stability globally. In this study, we conducted a comprehensive landslide susceptibility mapping (LSM) in the western region of Cameroon, focusing on Bafoussam and its surroundings. The integration of multi-criteria decision analysis models (AHP), statistical methods (Information Value IV, Shannon Entropy SE, Frequency Ratio FR), and machine learning algorithms (Naïve Bayes and Logistic Regression) provided a robust assessment of landslide risk. Our analysis, based on 54 recorded landslides, carefully selected Landslide Conditioning Factors (LCF), and influential parameters such as lithology, slope, altitude, and precipitation, resulted in susceptibility maps categorizing the area into five risk zones. The Spatial distribution shows the centre and northwestern regions as high-risk areas. Model sensitivity differences underscore the need for tailored LSM selection. Validation using the Area Under Curve/Receiver Operating Characteristics (AUC/ROC) method indicates the LR and NB methods have the highest accuracy (82.7% and 84.1%, respectively). Comparative analysis of landslide events in Gouaché, Sichuan, Souk Ahras, and Kekem reveals correlations between heavy rainfall and geological conditions. The study supplies valuable insights for decision-makers in landslide-prone areas, emphasizing the importance of integrating multiple methodologies for comprehensive risk assessment and management. [Display omitted] • A comprehensive landslide susceptibility mapping (LSM) was conducted in the west region of Cameroon, focusing on Bafoussam and its surroundings. • The study utilized a combination of Multi-Criteria Decision Analysis models, statistical methods, and machine learning algorithms for a robust assessment of landslide risk. • Susceptibility maps classified the area into five risk zones, with the centre and northwestern regions identified as high-risk areas. • Validation using the AUC/ROC method showed the Logistic Regression and Naïve Bayes methods had the highest accuracy (82.7% and 84.1%, respectively). • The study underscores the need for a tailored approach to LSM selection and the importance of integrating multiple methodologies for comprehensive risk assessment and management. [ABSTRACT FROM AUTHOR]

• Relationships among amyloid PET, MRI, and cognition are unclear in Alzheimer. • Deep-SHAP is proposed to investigate regional imaging vs. cognition relationships. • Deep-SHAP identified the modality-specific brain regions relevant to cognition. Mild cognitive impairment (MCI)/Alzheimer's disease (AD) is associated with cognitive decline beyond normal aging and linked to the alterations of brain volume quantified by magnetic resonance imaging (MRI) and amyloid-beta (Aβ) quantified by positron emission tomography (PET). Yet, the complex relationships between these regional imaging measures and cognition in MCI/AD remain unclear. Explainable artificial intelligence (AI) may uncover such relationships. We integrate the AI-based deep learning neural network and Shapley additive explanations (SHAP) approaches and introduce the Deep-SHAP method to investigate the multivariate relationships between regional imaging measures and cognition. After validating this approach on simulated data, we apply it to real experimental data from MCI/AD patients. Deep-SHAP significantly predicted cognition using simulated regional features and identified the ground-truth simulated regions as the most significant multivariate predictors. When applied to experimental MRI data, Deep-SHAP revealed that the insula, lateral occipital, medial frontal, temporal pole, and occipital fusiform gyrus are the primary contributors to global cognitive decline in MCI/AD. Furthermore, when applied to experimental amyloid Pittsburgh compound B (PiB)-PET data, Deep-SHAP identified the key brain regions for global cognitive decline in MCI/AD as the inferior temporal, parahippocampal, inferior frontal, supratemporal, and lateral frontal gray matter. Deep-SHAP method uncovered the multivariate relationships between regional brain features and cognition, offering insights into the most critical modality-specific brain regions involved in MCI/AD mechanisms. [ABSTRACT FROM AUTHOR]

• Hybrid models successfully predicted key performance metrics for antibody production in mammalian cells. • Such models allow predictions under varying process conditions, without the need to re-estimate parameters. • Performance could be predicted for a different product under similar process conditions with reasonable accuracy. • Algorithm choice did not significantly impact model performance, but computational requirements differed. Among the many approaches to build predictive models for bioprocesses, hybrid models have gained increased attention due to their ability to combine data driven approaches with process knowledge. This study develops and compares hybrid models that predict the performance of a mammalian CHO cell culture producing a therapeutic product. Three machine learning algorithms, MLP, Random Forest and XGBoost regressors are compared to understand the effect of algorithm choice on the rates predicted and overall performance of the hybrid model. When combined in series with mechanistic equations, all three algorithms could predict next day Viable Cell Density, titer and the cumulative amount of glucose consumed with low error, although decision tree-based algorithm required less computational time for training. The models performed well using a dataset for a new product, without further model training. Retraining the model on a subset of data from the new product improved the prediction accuracy of VCD and titer. The potential for application and further model developments are also outlined. [Display omitted] [ABSTRACT FROM AUTHOR]

This study investigates how urban configuration influences the distribution of heat, known as the Urban Heat Island (UHI) effect, in Shiraz, Iran. Several Machine Learning algorithms are employed to analyze Land Surface Temperature (LST) data across various land cover types, including built-up, soil, and vegetation. The analysis reveals that Deep Neural Networks (DNNs) and Extreme Gradient Boosting (XGB) models excel at predicting LST, outperforming other methods. These results highlight the significant impact of land use on LST patterns within the metropolitan regions. Furthermore, the study assesses the influence of specific configuration metrics within each land cover category. This allows researchers to pinpoint which urban morphology features most significantly affect LST. These insights can inform targeted interventions and management strategies implemented to mitigate heat and improve thermal comfort in specific areas of Shiraz. • Urban land use changes can raise surface temperatures (LST). XGB model predicted LST best out of six tested, followed by DNN. Morphology features (like green space size) affect LST differently in each area. [ABSTRACT FROM AUTHOR]