Machine Learning Approach Research Articles

Prediction of talent selection in elite male youth soccer across 7 seasons: A machine-learning approach

ABSTRACT This study aimed to investigate the relative importance of parameters from several domains associated to both selecting or de-selecting players with regards to the next age group within a professional German youth soccer academy across a 7-year period. Following a mixed-longitudinal approach, physical, physiological, psychological, skill-, health-, age-, and position-related parameters were collected from 409 male players (980 datapoints) from the U12 to U19 age groups. Supervised machine learning classifiers were used to predict the selection status regarding the next age group. The XGBoost models (ROC-AUC: 0.69, F1-Score: 0.84) revealed that physical and physiological (linear sprint, change-of-direction sprint, countermovement jump, aerobic speed reserve) as well as skill-related parameters (soccer-specific skill) were most important for being selected or de-selected regarding the next age group across the entire sample and all age groups. The majority of psychological parameters (motive structure, motive attention, motive competition, cognitive flexibility) were of medium importance. No clear pattern was observed for the health-, age-, and position-related parameters. Our study provides insights into key parameters for talent selection thereby contributing to an overall talent management strategy in highly trained youth soccer players. In particular, coaches and key stakeholders might focus on physical, physiological, and skill-related parameters.

Long-term courses in schizophrenia : Areview of current results and developments

Despite classical conceptions of schizophrenia as aprogressive illness with a high chronification, current long-term follow-up studies show a higher proportion of remission, possibly also a higher proportion of recovery than previously assumed. The heterogeneity of clinical courses is also reflected in different trajectories of cognitive and biological (e.g., imaging) variables, in which many of those affected show remission. Early clinical intervention during the first weeks and months following the onset of psychosis are decisive not only for early remission but also possibly for the long-term outcome. The treatment and reduction of somatic comorbidities are promising approaches in addition to a differentiated core treatment to positively influence the course of the illness even years after the disease. The identification of additional predictors, e.g., based on biological parameters, can together with machine learning approaches contribute to optimization of an individualized core treatment.

Modeling impacts of traffic, air pollution, and weather conditions on cardiopulmonary disease mortality.

Cardiopulmonary disease (CPD) is a leading cause of death worldwide. Increasing evidence shows that air pollution and exposure to weather conditions have important contributory roles. Understanding the interaction of these factors is difficult due to the complexity of the relationship between CPD, air pollution, and environmental factors. This paper uses regression models and machine learning approaches to explore these relationships, and investigate whether meteorological factors and air pollution have a synergistic effect on CPD. We use daily data from 2009-2018 from four cities representing the heterogenous climate conditions in Norway: the far north, the west coast, mid-Norway, and the south-east. We demonstrate the importance of the interaction between weather and air pollution associated with higher CPD mortality, as is exposure to air pollution in the form of particulate matter. This impact is seasonal. Traffic is also positively related to CPD mortality, which may be caused indirectly through increased pollution. We demonstrate that machine learning outperforms regression models in terms of the accuracy of predicting CPD mortality. The inclusion of rich lagged structures and interactions between environmental factors are both important but can lead to overfitting of traditional models; since these cities are not large cities by international standards, it is surprising that environmental factors have such obvious impacts on CPD mortality. CPD mortality shows a clear negative trend, implying an improvement in the public health situation.

Prediction on mechanical properties of engineered cementitious composites: An experimental and machine learning approach

AbstractThe adoption of engineered cementitious composites (ECC) has witnessed a notable surge in recent years, primarily because of their remarkable strain‐hardening behavior and other hardened properties. On the other hand, machine learning (ML) approaches have been widely employed to predict various properties in engineering applications by incorporating an ‘n’ number of inputs and target data. An ML aids in understanding the selection, properties, and blending ability of materials, thereby reducing the cost and duration of research. In this study, an ML technique based on an artificial neural network (ANN) is utilized to predict the mechanical properties of ECC with different binders. For this purpose, several parameters are collected from the present investigation and various literature sources, and the collected data are trained using three algorithms, namely, scaled conjugate gradient (SCG), Bayesian regularization (BR), and Levenberg–Marquardt (LM). The correlations and variations between the experimental and predicted outputs are analyzed. In addition, a comparison between the experimental results obtained by each investigator and the corresponding outputs predicted by the individual algorithms is highlighted. The LM algorithm achieved a mean regression value of 0.910 for the prediction of compressive strength, whereas the BR showed values of 0.908 and 0.852 for predicting the direct tensile and flexural properties of ECC, respectively. Furthermore, considering the standard benchmark, the proposed model exhibited a high correlation with the coefficient of determination (R2).

A systematic review of deep learning applications in database query execution

Modern database management systems (DBMS), primarily designed as general-purpose systems, face the challenging task of efficiently handling data from diverse sources for both analytical services and online transactional processing (OLTP). The volume of data has grown significantly, with distributions ranging from linear to highly skewed, sometimes involving very slow changes or rapid, intensive updates. Recent research in this field has been significantly influenced by advances in machine learning (ML), particularly deep learning (DL), and these developments have led to the application of various ML algorithms to enhance the efficiency of different parts of the query execution engine. While previous research studies were mostly focused on identifying drawbacks to individual components, such as the query optimizer, there is a notable lack of studies examining the applicability and effectiveness of various machine learning approaches across multiple aspects of the query execution engine. This article aims to provide a systematic review of approaches that apply deep learning models at various levels within the query execution engine. We categorize these approaches into three groups based on how such models are applied: improving performance of index structures and consequently data manipulation algorithms, query optimization tasks, and externally controlling query optimizers through parameter tuning. Furthermore, we discuss the key challenges associated with implementing deep learning algorithms in DBMS.

Water source, latrine type, and rainfall are associated with detection of non-optimal and enteric bacteria in the vaginal microbiome: a prospective observational cohort study nested within a cluster randomized controlled trial

BackgroundLess than one-third of sub-Saharan Africans have access to improved water sources. In US, Indian, and African studies, Bacterial vaginosis (BV) is increased among women with poor water, sanitation, and hygiene (WASH). We examined water source, sanitation (latrine type), and rainfall in relation to the vaginal microbiome (VMB).MethodsIn a cluster randomized controlled trial of menstrual cups and cash transfer, we measured the impact of cups on VMB via 16S rRNA gene amplicon sequencing in a subset of 436 adolescent girls. We analyzed how self-reported water source and latrine type at home related to VMB over 18-months, examining community state type I (CST-I, L. crispatus dominant) vs. other CST; alpha diversity; targeted taxa (coliform and other water-related pathogens); and non-targeted taxa via machine learning approaches. Mixed effects multivariable longitudinal models were adjusted for intervention arm, age, socioeconomic status, sexual activity, and cluster-level school WASH and rainfall (in millimeters).ResultsAdjusting for all covariates in all models: (1) the odds of CST-I were increased among participants with piped water (vs. pond), and decreased with traditional pit latrine vs. flush toilet. (2) Alpha diversity varied by water source and latrine type without consistent trends. (3) Coliform bacteria relative abundance (RA) was higher among participants with traditional pit or ventilated improved pit latrines vs. flush toilet, and higher among participants relying on stream vs. pond water. Streptococcus agalactiae RA was higher among participants with non-flush toilets, while Bacteroides fragilis RA was lower with non-flush toilets. (4) Key taxa from non-targeted analyses associated with water source and latrine type included typical vaginal bacteria, opportunistic pathogens, and urinary tract pathobionts. (6) Increased rainfall was associated with decreased odds of CST-I.Trial registrationClinicalTrials.gov NCT03051789, February 14, 2017.

An integrated machine learning approach delineates an entropic expansion mechanism for the binding of a small molecule to α-synuclein

The mis-folding and aggregation of intrinsically disordered proteins (IDPs) such as α-synuclein (αS) underlie the pathogenesis of various neurodegenerative disorders. However, targeting αS with small molecules faces challenges due to the lack of defined ligand-binding pockets in its disordered structure. Here, we implement a deep artificial neural network-based machine learning approach, which is able to statistically distinguish the fuzzy ensemble of conformational substates of αS in neat water from those in aqueous fasudil (small molecule of interest) solution. In particular, the presence of fasudil in the solvent either modulates pre-existing states of αS or gives rise to new conformational states of αS, akin to an ensemble-expansion mechanism. The ensembles display strong conformation-dependence in residue-wise interaction with the small molecule. A thermodynamic analysis indicates that small-molecule modulates the structural repertoire of αS by tuning protein backbone entropy, however entropy of the water remains unperturbed. Together, this study sheds light on the intricate interplay between small molecules and IDPs, offering insights into entropic modulation and ensemble expansion as key biophysical mechanisms driving potential therapeutics.

Romanian Fake News Detection Using Machine Learning and Transformer-Based Approaches

Nowadays, the consequence of quick access to information has lead to the spread of fake news, which has a strong damaging impact on democracy, justice, and public trust. Thus, it is crucial to analyze and evaluate detection methods for fake news. This paper focuses on the detection of Romanian fake news. In this study, we made a comparative analysis of machine learning algorithms and Transformer-based models on Romanian fake news detection using three datasets—FakeRom, NEW, and both FakeRom + NEW. The NEW dataset was build using a scrapping algorithm applied on the Veridica platform. Our approach uses the following machine learning models for detection: Naive Bayes (NB), Logistic Regression (LR), and Support Vector Machine (SVM). We also used two Transformer-based models—BERT-based-multilingual-cased and RoBERTa-large. The performance of the models was evaluated using various metrics: accuracy, precision, recall, and F1 score. The results revealed that the BERT model trained on the NEW dataset consistently achieved the highest performance metrics across all test sets, with 96.5%. Also, Support Vector Machine trained on NEW was another top performer, reaching a very good accuracy of 94.6% on the combined test set.

Integrative Stacking Machine Learning Model for Small Cell Lung Cancer Prediction Using Metabolomics Profiling

Background: Small cell lung cancer (SCLC) is an extremely aggressive form of lung cancer, characterized by rapid progression and poor survival rates. Despite the importance of early diagnosis, the current diagnostic techniques are invasive and restricted. Methods: This study presents a novel stacking-based ensemble machine learning approach for classifying small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) using metabolomics data. The analysis included 191 SCLC cases, 173 NSCLC cases, and 97 healthy controls. Feature selection techniques identified significant metabolites, with positive ions proving more relevant. Results: For multi-class classification (control, SCLC, NSCLC), the stacking ensemble achieved 85.03% accuracy and 92.47 AUC using Support Vector Machine (SVM). Binary classification (SCLC vs. NSCLC) further improved performance, with ExtraTreesClassifier reaching 88.19% accuracy and 92.65 AUC. SHapley Additive exPlanations (SHAP) analysis revealed key metabolites like benzoic acid, DL-lactate, and L-arginine as significant predictors. Conclusions: The stacking ensemble approach effectively leverages multiple classifiers to enhance overall predictive performance. The proposed model effectively captures the complementary strengths of different classifiers, enhancing the detection of SCLC and NSCLC. This work accentuates the potential of combining metabolomics with advanced machine learning for non-invasive early lung cancer subtype detection, offering an alternative to conventional biopsy methods.

Comparing and Optimizing Four Machine Learning Approaches to Radar-Based Quantitative Precipitation Estimation

To improve radar-based quantitative precipitation estimation (QPE) methods, this study investigated the relationship between radar reflectivity (Z) and hourly rainfall intensity (R) using data from 289 precipitation events in Shanghai between September 2020 and March 2024. Two Z-R relationship models were compared in terms of their fitting performance: Z = 270.81 R1.09 (empirically fitted relationship) and Z = 300 R1.4 (standard relationship). The results show that the Z = 270.81 R1.09 model outperforms the Z = 300 R1.4 model in terms of fitting accuracy. Specifically, the Z = 270.81 R1.09 model more effectively captures the nonlinear relationship between radar reflectivity and rainfall intensity, with a higher degree of agreement between the fitted curve and the observed data points. This model demonstrated superior performance across all 289 precipitation events. This study evaluated the performance of four machine learning approaches while incorporating five meteorological features: specific differential phase shift (KDP), echo-top height (ET), vertical liquid water content (VIL), differential reflectivity (ZDR), and correlation coefficient (CC). Nine QPE models were constructed using these inputs. The key findings are as follows: (1) For models with a single-variable input, the KAN deep learning model outperformed Random Forest, Gradient Boosting Decision Trees, Support Vector Machines, and the traditional Z-R relationship. (2) When six features were used as inputs, the accuracy of the machine learning models improved significantly, with the KAN deep learning model outperforming other machine learning methods. Compared to using only radar reflectivity, the KAN deep learning model reduced the MRE by 20.78%, MAE by 4.07%, and RMSE by 12.74%, while increasing the coefficient of determination (R2) by 18.74%. (3) The integration of multiple meteorological features and machine learning optimization significantly enhanced QPE accuracy, with the KAN deep learning model performing best under varying meteorological conditions. This approach offers a promising method for improving radar-based QPE, particularly considering seasonal, weather system, and precipitation stage differentiation.

Transformers deep learning models for missing data imputation: an application of the ReMasker model on a psychometric scale

IntroductionMissing data in psychometric research presents a substantial challenge, impacting the reliability and validity of study outcomes. Various factors contribute to this issue, including participant non-response, dropout, or technical errors during data collection. Traditional methods like mean imputation or regression, commonly used to handle missing data, rely upon assumptions that may not hold on psychological data and can lead to distorted results.MethodsThis study aims to evaluate the effectiveness of transformer-based deep learning for missing data imputation, comparing ReMasker, a masking autoencoding transformer model, with conventional imputation techniques (mean and median imputation, Expectation–Maximization algorithm) and machine learning approaches (K-nearest neighbors, MissForest, and an Artificial Neural Network). A psychometric dataset from the COVID distress repository was used, with imputation performance assessed through the Root Mean Squared Error (RMSE) between the original and imputed data matrices.ResultsResults indicate that machine learning techniques, particularly ReMasker, achieve superior performance in terms of reconstruction error compared to conventional imputation techniques across all tested scenarios.DiscussionThis finding underscores the potential of transformer-based models to provide robust imputation in psychometric research, enhancing data integrity and generalizability.

Sustainable integration of artificial intelligence and machine learning approaches within the African infectious disease vaccine research and development ecosystem

Artificial Intelligence and Machine Learning (AI/ML) techniques, including reverse vaccinology and predictive models, have already been applied for developing vaccine candidates for COVID-19, HIV, and Hepatitis, streamlining the vaccine development lifecycle from discovery to deployment. The application of AI and ML technologies for improving heath interventions, including drug discovery and clinical development, are expanding across Africa, particularly in South Africa, Kenya, and Nigeria. Further initiatives are required however to expand AI/ML capabilities across the continent to ensure the development of a sustainable ecosystem including enhancing the requisite knowledge base, fostering collaboration between stakeholders, ensuring robust regulatory and ethical frameworks and investment in requisite infrastructure.

Deciphering key roles of B cells in prognostication and tailored therapeutic strategies for lung adenocarcinoma: a multi-omics and machine learning approach towards predictive, preventive, and personalized treatment strategies

Deciphering key roles of B cells in prognostication and tailored therapeutic strategies for lung adenocarcinoma: a multi-omics and machine learning approach towards predictive, preventive, and personalized treatment strategies

Using Machine Learning Approaches to Enhance Heatwave Measurement for Vulnerability Assessment and Timely Management of Heat-related Health Services

Climate change is one of the most critical challenges facing Australia and the global community today. Data from the Australian Bureau of Meteorology (BoM) indicates that Australia has been experiencing rising temperatures, particularly since the late 20th century. The frequency, duration, and intensity of heatwaves are projected to continue increasing . Since national records began in 1910, Australia has warmed by an average of 1.47°C (±0.24°C), with the highest official temperature recorded at 50.7 degrees Celsius in Onslow, Western Australia (WA), on January 13, 2022. Furthermore, a recent unprecedented high temperature of +41.6°C was recorded during winter on August 26, 2024, in Yampi Sound, WA. Among all natural disasters in Australia, heatwave (HW) represents a leading silent killer and pose a significant public health threat. However, innovative methods for assessing vulnerability for HW-related health services remain limited.

Optimizing Stabilization of Contaminated Mining Sludge: A Machine Learning Approach to Predict Strength and Heavy Metal Leaching

Optimizing Stabilization of Contaminated Mining Sludge: A Machine Learning Approach to Predict Strength and Heavy Metal Leaching

Bridging the Gap: Unveiling the Potential of Tanzania’s SMEs through VAT Insights

Abstract The collection of value added tax (VAT) is essential for achieving domestic revenue objectives, yet VAT gap estimation is rarely performed in developing countries. This study utilizes innovative tax declaration and audit data to estimate VAT misreporting in Tanzania, applying a machine learning approach to predict evasion in unaudited firms and periods. We measure the underreporting component of the compliance gap, quantifying potential revenue losses due to inaccurate reporting. Our findings indicate that firms often avoid excessive audits, leading to increased evasion, with those firms showing the largest VAT gaps. We estimate a 62% VAT gap among small and medium-sized enterprises in Tanzania. Finally, we present a cost-benefit ratio, suggesting that while auditing sectors with large VAT gaps is cost-effective, those contributing significantly to the overall VAT gap should also be prioritized for revenue generation. (JEL codes: H25, H26, and H32)

Advancing Grapevine Variety Identification: A Systematic Review of Deep Learning and Machine Learning Approaches

The Eurasian grapevine (Vitis vinifera L.) is one of the most extensively cultivated horticultural crop worldwide, with significant economic relevance, particularly in wine production. Accurate grapevine variety identification is essential for ensuring product authenticity, quality control, and regulatory compliance. Traditional identification methods have inherent limitations limitations; ampelography is subjective and dependent on skilled experts, while molecular analysis is costly and time-consuming. To address these challenges, recent research has focused on applying deep learning (DL) and machine learning (ML) techniques for grapevine variety identification. This study systematically analyses 37 recent studies that employed DL and ML models for this purpose. The objective is to provide a detailed analysis of classification pipelines, highlighting the strengths and limitations of each approach. Most studies use DL models trained on leaf images captured in controlled environments at distances of up to 1.2 m. However, these studies often fail to address practical challenges, such as the inclusion of a broader range of grapevine varieties, using data directly acquired in the vineyards, and the evaluation of models under adverse conditions. This review also suggests potential directions for advancing research in this field.

Potential shared mechanisms in atopic dermatitis and type 2 diabetes identified via transcriptomic and machine learning approaches

Although atopic dermatitis (AD) and type 2 diabetes mellitus (T2DM) may appear clinically and pathophysiologically unrelated, AD is a common skin disease characterized by chronic inflammation and skin barrier dysfunction, whereas T2DM is a metabolic disorder marked by hyperglycemia and chronic inflammation, which further exacerbates insulin resistance (IR) through the release of systemic inflammatory factors. Despite their apparent differences, the molecular mechanisms shared between AD and T2DM remain relatively unexplored. In this study, we integrated transcriptomic data from both AD and T2DM using differential gene expression analyses (DEGs), gene set variation analysis (GSVA), and machine learning algorithms to uncover common features of these diseases. We identified several characteristic genes, including LTF, LTB4R, and CCR1, which are significantly upregulated in both conditions and may serve as potential biomarkers. Furthermore, virtual screening revealed that Dioscin, Camptothecin, and Albamycin exhibit strong affinity for the CCR1 binding site, indicating their potential as therapeutic candidates. In summary, this study elucidates the shared molecular mechanisms of AD and T2DM and introduces new potential targets and drugs for the diagnosis and treatment of these diseases.

Where to Build Food Banks: A Machine Learning Approach

Where to Build Food Banks: A Machine Learning Approach

EnConv: enhanced CNN for leaf disease classification

Detecting leaf diseases in plants is essential to maintain crop yield and market value. Machine learning has shown promise in detecting these diseases as it can group data into predetermined categories after examining it from various angles. However, machine learning models require a thorough knowledge of plant diseases, and processing time can be lengthy. This study proposes an enhanced convolutional neural network that utilizes depthwise separable convolution and inverted residual blocks to detect leaf diseases in plants. The model considers the morphological properties and characteristics of the plant leaves, including color, intensity, and size, to categorize the data. The proposed model outperforms traditional machine learning approaches and deep learning models, achieving an accuracy of 99.87% for 39 classes of different plants such as tomato, corn, apple, potato, and more. To further improve the model, global average pooling was used in place of the flatten layer. Overall, this study presents a promising approach to detect leaf diseases in plants using an enhanced convolutional neural network with depthwise separable convolution and inverted residual blocks. The results show the potential benefits of using this model in agriculture to improve the early detection of plant diseases and maintain crop yield and market value.