Search Results (701)

This paper proposes a methodology for implementing a custom-developed explainability model for credit rating using behavioral data registered during the lifecycle of the borrowing that can replicate the score given by the regulatory model for the solidarity economy in Colombia. The methodology integrates continuous behavioral and financial variables from over 17,000 real credit histories into predictive models based on ridge regression, decision trees, random forests, XGBoost, and LightGBM. The models were trained and evaluated using cross-validation and RMSE metrics. LightGBM emerged as the most accurate model, effectively capturing nonlinear credit behavior patterns. To ensure interpretability, SHAP was used to identify the contribution of each feature to the model predictions. The presented model using LightGBM predicted the credit risk assessment in accordance with the regulatory model used by the Colombian Superintendence of the Solidarity Economy, with a root-mean-square error of 0.272 and an R² score of 0.99. We propose an alternative framework using explainable machine learning models aligned with the internal ratings-based approach under Basel II. Our model integrates variables collected throughout the borrowing lifecycle, offering a more comprehensive perspective than the regulatory model. While the regulatory framework adjusts itself generically to national regulations, our approach explicitly accounts for borrower-specific dynamics. Full article

The Directed Energy Deposition (DED) process has demonstrated high efficiency in manufacturing steel parts with complex geometries and superior capabilities. Understanding the complex interplays of alloy compositions, cooling rates, grain sizes, thermal histories, and mechanical properties remains a significant challenge during DED processing. Interpretable and data-driven modeling has proven effective in tackling this challenge, as machine learning (ML) algorithms continue to advance in capturing complex property structural relationships. However, accurately predicting the prime mechanical properties, including ultimate tensile strength (UTS), yield strength (YS), and hardness value (HV), remains a challenging task due to the complex and non-linear relationships among process parameters, material constituents, grain size, cooling rates, and thermal history. This study introduces an ML model capable of accurately predicting the UTS, YS, and HV of a material dataset comprising 4900 simulation analyses generated using the “JMatPro” software, with input parameters including material compositions, grain size, cooling rates, and temperature, all of which are relevant to DED-processed low-alloy steels. Subsequently, an ML model is developed using the generated dataset. The proposed framework incorporates a physics-based DED-specific feature that leverages “JMatPro” simulations to extract key input parameters such as material composition, grain size, cooling rate, and thermal properties relevant to mechanical behavior. This approach integrates a suite of flexible ML algorithms along with customized evaluation metrics to form a robust foundation to predict mechanical properties. In parallel, explicit data-driven models are constructed using Multivariable Linear Regression (MVLR), Polynomial Regression (PR), Multi-Layer Perceptron Regressor (MLPR), XGBoost, and classification models to provide transparent and analytical insight into the mechanical property predictions of DED-processed low-alloy steels. Full article

Background/Objectives: Gallstone disease, a prevalent and costly digestive system disorder, is influenced by multifactorial risk factors, some of which may predispose to malignancy. This study aims to evaluate the association between gallstone disease and malignancy using advanced machine learning (ML) algorithms. Methods: A dataset comprising approximately 1000 patients was analyzed, employing six ML methods: random forests (RFs), support vector machines (SVMs), multi-layer perceptron (MLP), MLP with PyTorch 2.3.1 (MLP_PT), naive Bayes (NB), and Tabular Prior-data Fitted Network (TabPFN). Comparative performance was assessed using Pearson correlation, sensitivity, specificity, Kappa, receiver operating characteristic (ROC), area under curve (AUC), and accuracy metrics. Results: Our results revealed that age, body mass index (BMI), and history of HRT were the most significant predictors of malignancy. Among the ML models, TabPFN emerged as the most effective, achieving superior performance across multiple evaluation criteria. Conclusions: This study highlights the potential of leveraging cutting-edge ML methodologies to uncover complex relationships in clinical datasets, offering a novel perspective on gallstone-related malignancy. By identifying critical risk factors and demonstrating the efficacy of TabPFN, this research provides actionable insights for predictive modeling and personalized patient management in clinical practice. Full article

Urban centers shaped by industrial histories often exhibit complex patterns of land cover change that are not well-captured by standard classification techniques. This study investigates post-industrial urban change in Ponca City, Oklahoma, using remote sensing, unsupervised machine learning, and socioeconomic contextualization. Using a Jupyter Notebook version 7.0.8 environment for Python libraries, Landsat imagery from 1990 to 2020 was analyzed to detect shifts in land cover patterns across a relatively small, heterogeneous landscape. Principal component analysis (PCA) was applied to reduce dimensionality and enhance pixel distinction across multiband reflectance data. Socioeconomic data and historical context were incorporated to interpret changes in land use alongside patterns of industrial reduction and urban redevelopment. Results revealed changes in five distinct land cover classes of urban, vegetative, and industrial land uses, with observable trends aligning with key periods of economic and infrastructural transition. The trends also aligned with socioeconomic changes of the city, with a larger reduction in industrial and commercial land cover than in residential and vegetation cover types. These findings demonstrate the utility of machine learning classification in small-scale, heterogeneous environments and provide a replicable methodological framework for smaller city municipalities to monitor urban change. Full article

Background: A short cervix in the second trimester significantly increases preterm birth risk, yet no reliable first-trimester prediction method exists. Current guidelines lack consensus on which women should undergo transvaginal ultrasound (TVUS) screening for cost-effective prevention. Therefore, it is vital to establish a highly accurate and economical method for use in the early stages of pregnancy to predict short cervix in mid-pregnancy. Methods: A total of 1480 pregnant women with singleton pregnancies and at least one risk factor for spontaneous preterm birth (<37 weeks) were recruited from January 2020 to December 2020 at the Shanghai First Maternity and Infant Hospital, Tongji University School of Medicine. Cervical length was assessed at 20–24 weeks of gestation, with a short cervix defined as <25 mm. Feature selection employed tree models, regularization, and recursive feature elimination (RFE). Seven machine learning models (logistic regression, linear discriminant analysis, k-nearest neighbors, support vector machine, decision tree, random forest, XGBoost) were trained to predict mid-trimester short cervix. The XGBoost model—an ensemble method leveraging sequential decision trees—was analyzed using Shapley Additive Explanation (SHAP) values to assess feature importance, revealing consistent associations between clinical predictors and outcomes that align with known clinical patterns. Results: Among 1480 participants, 376 (25.4%) developed mid-trimester short cervix. The XGBoost-based prediction model demonstrated high predictive performance in the training set (Recall = 0.838, F1 score = 0.848), test set (Recall = 0.850, F1 score = 0.910), and an independent dataset collected in January 2025 (Recall = 0.708, F1 score = 0.791), with SHAP analysis revealing pre-pregnancy BMI as the strongest predictor, followed by second-trimester pregnancy loss history, peripheral blood leukocyte count (WBC), and positive vaginal microbiological culture results (≥10⁵ CFU/mL, measured between 11⁺⁰ and 13⁺⁶ weeks). Conclusions: The XGBoost model accurately predicts mid-trimester short cervix using first-trimester clinical data, providing a 6-week window for targeted interventions before the 20–24-week gestational assessment. This early prediction could help guide timely preventive measures, potentially reducing the risk of spontaneous preterm birth (sPTB). Full article

The objective of this study is to classify the types of boxing punches using machine learning algorithms that processed skin-strain time history measurements from a self-adhesive, elastic fabric, wearable sensor called Motion Tape. A human participant study was designed to capture movements during boxing training. Subjects were asked to perform multiple sets of punches during the entire test, which consisted of jabs and hooks with and without striking a heavy bag. The collected Motion Tape data was used to train and compare time series classification algorithms to identify the types of punches performed and associated conditions. The results demonstrated that Motion Tape, in combination with machine learning techniques, could effectively classify different punch types based on skin-strain measurements. These findings highlighted the potential of the system as an effective tool for human performance analysis in sports and biomechanics applications. Full article

Background/Objectives: Preeclampsia (PE) is a major cause of maternal and perinatal morbidity and mortality, particularly in low- and middle-income countries. Early-onset PE (EOP) and late-onset PE (LOP) are distinct clinical entities with differing pathophysiological mechanisms and prognoses. However, few studies have explored differential risk factors for EOP and LOP in Latin American populations. This study aimed to identify and assess clinical risk factors for predicting EOP and LOP in a cohort of pregnant women from Bogotá, Colombia, using traditional statistics and machine learning (ML). Methods: A cross-sectional observational study was conducted on 190 pregnant women diagnosed with PE (EOP = 80, LOP = 110) at a tertiary hospital in Bogotá between 2017 and 2018. Risk factors and perinatal outcomes were collected via structured interviews and clinical records. Traditional statistical analyses were performed to compare the study groups and identify associations between risk factors and outcomes. Eleven ML techniques were used to train and externally validate predictive models for PE subtype and secondary outcomes, incorporating permutation-based feature importance to enhance interpretability. Results: EOP was significantly associated with higher maternal education and history of hypertension, while LOP was linked to a higher prevalence of allergic history. The best-performing ML model for predicting PE subtype was linear discriminant analysis (recall = 0.71), with top predictors including education level, family history of perinatal death, number of sexual partners, primipaternity, and family history of hypertension. Conclusions: EOP and LOP exhibit distinct clinical profiles in this cohort. The combination of traditional statistics with ML may improve early risk stratification and support context-specific prenatal care strategies in similar settings. Full article

Accurately predicting patient clinical outcomes is a complex task that requires integrating diverse factors, including individual characteristics, treatment histories, and environmental influences. This challenge is further exacerbated by missing data and inconsistent data quality, which often hinder the effectiveness of traditional single-task learning (STL) models. Multi-Task Learning (MTL) has emerged as a promising paradigm to address these limitations by jointly modeling related prediction tasks and leveraging shared information. In this study, we proposed MTLNFM, a multi-task learning framework built upon Neural Factorization Machines, to jointly predict patient clinical outcomes on a cohort of 2001 ICU patients. We designed a preprocessing strategy in the framework that transforms missing values into informative representations, mitigating the impact of sparsity and noise in clinical data. We leveraged the shared representation layers, composed of a factorization machine and dense neural layers that can capture high-order feature interactions and facilitate knowledge sharing across tasks for the prediction. We conducted extensive comparative experiments, demonstrating that MTLNFM outperforms STL baselines across all three tasks (i.e., frailty status, hospital length of stay and mortality prediction), achieving AUROC scores of 0.7514, 0.6722, and 0.7754, respectively. A detailed case analysis further revealed that MTLNFM effectively integrates both task-specific and shared representations, resulting in more robust and realistic predictions aligned with actual patient outcome distributions. Overall, our findings suggest that MTLNFM is a promising and practical solution for clinical outcome prediction, particularly in settings with limited or incomplete data, and can support more informed clinical decision-making and resource planning. Full article

Neuromuscular hip dysplasia (NHD) is a common deformity in children with cerebral palsy (CP). Although some predictive factors of NHD are known, the prediction of NHD is in its infancy. We present a Clinical Decision Support System (CDSS) designed to calculate the probability of developing NHD in children with CP. The system utilizes an ensemble of three machine learning (ML) algorithms: Neural Network (NN), Support Vector Machine (SVM), and Logistic Regression (LR). The development and evaluation of the CDSS followed the DECIDE-AI guidelines for AI-driven clinical decision support tools. The ensemble was trained on a data series from 182 subjects. Inclusion criteria were age between 12 and 18 years and diagnosis of CP from two specialized units. Clinical and functional data were collected prospectively between 2005 and 2023, and then analyzed in a cross-sectional study. Accuracy and area under the receiver operating characteristic (AUROC) were calculated for each method. Best logistic regression scores highlighted history of previous orthopedic surgery (p = 0.001), poor motor function (p = 0.004), truncal tone disorder (p = 0.008), scoliosis (p = 0.031), number of affected limbs (p = 0.05), and epilepsy (p = 0.05) as predictors of NHD. Both accuracy and AUROC were highest for NN, 83.7% and 0.92, respectively. The novelty of this study lies in the development of an efficient Clinical Decision Support System (CDSS) prototype, specifically designed to predict future outcomes of neuromuscular hip dysplasia (NHD) in patients with cerebral palsy (CP) using clinical data. The proposed system, PredictMed-CDSS, demonstrated strong predictive performance for estimating the probability of NHD development in children with CP, with the highest accuracy achieved using neural networks (NN). PredictMed-CDSS has the potential to assist clinicians in anticipating the need for early interventions and preventive strategies in the management of NHD among CP patients. Full article

The integration of machine learning (ML) into industrial automation is fundamentally reshaping how manufacturing systems are monitored, inspected, and optimized. By applying machine learning to real-time sensor data and operational histories, advanced models enable proactive fault prediction, intelligent inspection, and dynamic process control—directly enhancing system reliability, product quality, and efficiency. This review explores the transformative role of ML across three key domains: Predictive Maintenance (PdM), Quality Control (QC), and Process Optimization (PO). It also analyzes how Digital Twin (DT) and Edge AI technologies are expanding the practical impact of ML in these areas. Our analysis reveals a marked rise in deep learning, especially convolutional and recurrent architectures, with a growing shift toward real-time, edge-based deployment. The paper also catalogs the datasets used, the tools and sensors employed for data collection, and the industrial software platforms supporting ML deployment in practice. This review not only maps the current research terrain but also highlights emerging opportunities in self-learning systems, federated architectures, explainable AI, and themes such as self-adaptive control, collaborative intelligence, and autonomous defect diagnosis—indicating that ML is poised to become deeply embedded across the full spectrum of industrial operations in the coming years. Full article

In high geothermal tunnels (>28 °C), curing temperature critically affects early-age concrete mechanics and durability. Uniaxial compression tests under six curing conditions, combined with CT scanning and machine learning-based crack analysis, were used to evaluate the impacts of curing age, temperature, and fiber content. The test results indicate that concrete exhibits optimal development of mechanical properties under ambient temperature conditions. Specifically, the elastic modulus increased by 33.85% with age in the room-temperature group (RT), by 23.35% in the fiber group (F), and decreased by 26.75% in the varying-temperature group (VT). A Weibull statistical damage-based constitutive model aligned strongly with the experimental data (R² > 0.99). Fractal analysis of CT-derived cracks revealed clear fractal characteristics in the log(Nr)–log(r) curves, demonstrating internal damage mechanisms under different thermal histories. Full article

This study developed a model to predict ships with high deficiency risk under Port State Control (PSC) through machine learning techniques, particularly the Random Forest algorithm. The study utilized actual ship inspection data from the Port of Singapore, comprehensively considering various operational and safety indicators of ships, including but not limited to flag state, ship age, past deficiencies, and detention history. By analyzing these factors in depth, this research enhances the efficiency and accuracy of PSC inspections, provides decision support for port authorities, and offers strategic guidance for shipping companies to comply with international safety standards. During the research process, I first conducted detailed data preprocessing, including data cleaning and feature selection, to ensure the effectiveness of model training. Using the Random Forest algorithm, I identified key factors influencing the detention risk of ships and established a risk prediction model accordingly. The model validation results indicated that factors such as ship age, tonnage, company performance, and flag state significantly affect whether a ship exhibits a high deficiency rate. In addition, this study explored the potential and limitations of applying the Random Forest model in predicting high deficiency risk under PSC, and proposed future research directions, including further model optimization and the development of real-time prediction systems. By achieving these goals, I hope to provide valuable experience for other global shipping hubs, promote higher international maritime safety standards, and contribute to the sustainable development of the global shipping industry. This research not only highlights the importance of machine learning in the maritime domain but also demonstrates the potential of data-driven decision-making in improving ship safety management and port inspection efficiency. It is hoped that this study will inspire more maritime practitioners and researchers to explore advanced data analytics techniques to address the increasingly complex challenges of global shipping. Full article

This work focuses on the possibilities of generating synthetic data for machine learning in hazard prediction in area-based speed monitoring systems. The purpose of the research conducted was to develop a methodology for generating realistic synthetic data to support the design of a continuous vehicle speed monitoring system to minimize the risk of traffic accidents caused by speeding. The SUMO traffic simulator was used to model driver behavior in the analyzed area and within a given road network. Data from OpenStreetMap and field measurements from over a dozen speed detectors were integrated. Preliminary tests were carried out to record vehicle speeds. Based on these data, several simulation scenarios were run and compared to real-world observations using average speed, the percentage of speed limit violations, root mean square error (RMSE), and percentage compliance. A new metric, the Combined Speed Accuracy Score (CSAS), has been introduced to assess the consistency of simulation results with real-world data. For this study, a basic hazard prediction model was developed using LoRaWAN sensor network data and environmental contextual variables, including time, weather, location, and accident history. The research results in a method for evaluating and selecting the simulation scenario that best represents reality and drivers’ propensities to exceed speed limits. The results and findings demonstrate that it is possible to produce synthetic data with a level of agreement exceeding 90% with real data. Thus, it was shown that it is possible to generate synthetic data for machine learning in hazard prediction for area-based speed control systems using traffic simulators. Full article

Artificial intelligence (AI) has emerged as a significant driving force in the development of technology and industry. It is also integrated with music as music AI in music generation and analysis. It originated from early algorithmic composition techniques in the mid-20th century. Recent advancements in machine learning and neural networks have enabled innovative music generation and exploration. This article surveys the development history and technical route of music AI, analyzes the current status and limitations of music artificial intelligence across various areas, including music generation and composition, rehabilitation and treatment, as well as education and learning. It reveals that music AI has become a promising creator in the field of music generation. The influence of music AI on the music industry and the challenges it encounters are explored. Additionally, an emotional music generation system driven by multimodal signals is proposed. Although music artificial intelligence technology still needs to be further improved, with the continuous breakthroughs in technology, it will have a more profound impact on all areas of music. Full article

The deformation behavior of polymers is influenced not only by service conditions such as temperature and the strain rate but also significantly by the formation process. However, existing simulation frameworks typically treat injection molding and the in-service mechanical response separately, making it difficult to capture the impact of the thermal history on large deformation behavior. In this study, the deformation behavior of injection-molded polycarbonate (PC) was investigated by accounting for its thermal history during formation, achieved through combined experimental characterization and constitutive modeling. PC specimens were prepared via injection molding followed by annealing at different molding/annealing temperatures and durations. Uniaxial tensile tests were conducted using a Zwick universal testing machine at strain rates of 10⁻³–10⁻¹ s⁻¹ and temperatures ranging from 293 K to 353 K to obtain stress–strain curves. The effects of the strain rate, testing temperature, and annealing conditions were thoroughly examined. Building upon a previously proposed phenomenological model, a new constitutive framework incorporating thermal history effects during formation was developed to characterize the large deformation behavior of PC. This model was implemented in ABAQUS/Explicit using a user-defined material subroutine. Predicted stress–strain curves exhibit excellent agreement with the experimental data, accurately reproducing elastic behavior, yield phenomena, and strain-softening and strain-hardening stages. Full article