Search Results (571)

Tumor mutational burden (TMB) is a key pan-cancer biomarker for immunotherapy selection, but its routine assessment by whole-exome sequencing (WES) or large next-generation sequencing (NGS) panels is costly, time-consuming, and constrained by tissue and DNA quality. In parallel, advances in computational pathology have enabled deep learning models to infer molecular biomarkers directly from hematoxylin and eosin (H&E) whole-slide images (WSIs), raising the prospect of a purely digital assay for TMB. In this comprehensive review, we surveyed PubMed and Scopus (2015–2025) to identify original studies that applied deep learning directly to H&E WSIs of human solid tumors for TMB estimation. Across the 17 eligible studies, deep learning models have been applied to predict TMB from H&E WSIs in a variety of tumors, achieving moderate to good discrimination for TMB-high versus TMB-low status. Multimodal architectures tended to outperform conventional CNN-based pipelines. However, heterogeneity in TMB cut-offs, small and imbalanced cohorts, limited external validation, and the black-box nature of these models limit clinical translation. Full article

In the world of e-commerce, ensuring customer satisfaction and retention depends on delivering an optimal user experience. As the primary point of contact between businesses and consumers, a user interface’s success hinges on personalized human–computer interaction. The goal of this paper is to introduce the concept of a self-adaptive multi-variant user interface based on a novel application of a three-way decision-making model, which allows for “accept”, “reject”, or “delay” decisions on UI changes. The proposed framework enables the delivery of a multi-variant e-commerce user interface. It leverages human-centered machine learning to identify homogeneous groups of customers for whom a layout tailored to their behavior can be offered. The functionality of the solution was verified through pilot implementation and experimental studies. The results positively validated the three-way decision algorithm and highlighted clear directions for its refinement. The primary contribution of this work is the novel adaptation of the three-way decision model to create an automated framework for e-commerce UI personalization, moving beyond the limitations of traditional binary A/B testing. This study demonstrates the practical feasibility of using a self-adaptive, multi-variant interface to significantly improve user experience and key business metrics. These results confirm the feasibility and effectiveness of using self-adaptive e-commerce interfaces to improve the user experience. The proposed framework represents a promising solution to the challenges posed by static interfaces and demonstrates the potential for wider application in the e-commerce domain and beyond. Full article

This study aims to predict Turkish consumer preferences for smart tablets on e-commerce platforms, focusing on consumer behavior in a developing country context. Key product attributes—such as processor speed, screen size, internal storage capacity, display resolution, RAM, processor core count, and battery capacity—were collected from major e-commerce websites in Turkey. Data analysis indicated that consumers predominantly prefer tablets with processor speeds between 1–3 GHz, internal storage capacities of 32–64 GB, 2–3 GB of RAM, screen sizes of 7–11 inches, and battery capacities between 5001–8000 mAh. To predict the most preferred tablet configurations, Artificial Neural Networks (ANNs), Deep Neural Networks (DNNs), and Random Forest (RF) models were developed and evaluated. Among these, the ANN model achieved the highest prediction accuracy, particularly regarding RAM preferences. The findings contribute to the growing body of research on consumer behavior modeling in emerging markets and may assist manufacturers and marketers in shaping strategic decisions related to product development and online retail strategies. Full article

This research explores the optimization of mechanical properties and predictive modeling of polylactic acid (PLA) reinforced with boron nitride nanoplatelets (BNNPs) using data-driven machine learning (ML) models. PLA-BNNP composites were fabricated through injection molding, with a focus on how key processing parameters influence their mechanical performance. A Taguchi L27 orthogonal array was applied to assess the effects of BNNP composition (0.02 wt.% and 0.04 wt.%), injection temperature (135–155 °C), injection speed (50–70 mm/s), and pressure (30–50 bar) on properties such as tensile strength, Young’s modulus, and hardness. The results indicated that a 0.04 wt.% BNNP loading improved tensile strength, Young’s modulus, and hardness by 18.6%, 32.7%, and 20.5%, respectively, compared to pure PLA. Taguchi analysis highlighted that higher BNNP concentrations, along with optimal injection temperatures, improved all mechanical properties, although excessive temperatures compromised tensile strength and modulus, while enhancing hardness. Analysis of variance (ANOVA) revealed that injection temperature was the dominant factor for tensile strength (68.88%) and Young’s modulus (86.39%), while BNNP composition played a more significant role in influencing hardness (78.83%). Predictive models were built using machine learning (ML) models such as Random Forest Regression (RFR), Gradient Boosting Regression (GBR), and Extreme Gradient Boosting (XGBoost). Among the ML models, XGBoost demonstrated the highest predictive accuracy, achieving R² values above 98% for tensile strength, 92–93% for Young’s modulus, and 96% for hardness, with low error metrics i.e., Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE). These findings underscore the potential of using BNNP reinforcement and machine learning-driven property prediction to enhance PLA nanocomposites’ mechanical performance, making them viable for applications in lightweight packaging, biomedical implants, consumer electronics, and automotive components, offering sustainable alternatives to petroleum-based plastics. Full article

E-commerce retailers bear substantial additional costs arising from high product return rates due to lenient return policies and consumers’ impulsive purchasing. This study aims to accurately predict product return behavior before payment, supporting proactive return management and reducing potential losses. Based on the Graph Transformer, we proposed a novel return prediction model, Returnformer, which focuses on capturing user–product connections represented in topological structures of bipartite graphs. The Returnformer first integrates global topological embeddings into original node features to alleviate structural information loss caused by graph partitioning. It then employs a Graph Transformer to capture long-range user–item dependencies within local subgraphs. In addition, a graph-level attention mechanism is introduced to facilitate the propagation of global return patterns across different subgraphs. Experiments on a real-world e-commerce dataset show that the Returnformer outperforms four machine learning models in terms of prediction accuracy, demonstrating superior performance compared to the state-of-the-art models. The proposed model enables retailers to identify potential return risks prior to payment, thereby supporting timely and proactive preventive interventions. Full article

Online review images shape consumer perceptions by offering visual cues of product quality and use. Existing studies focus on aesthetics or object presence but overlook the functional balance among image types. This study introduces the Holistic Image Proportion (HIP)—the ratio of holistic to detailed review images—as a key determinant of visual information completeness. Using deep learning (ResNet-101) to classify over 240,000 images from 4450 clothing products, we find an inverted U-shaped relationship between HIP and sales: a balanced mix (HIP ≈ 0.5) maximizes performance. A follow-up experiment confirms that balanced image composition enhances perceived completeness, which fully mediates its effect on purchase intention. Review sentiment further moderates this relationship, amplifying the effect under positive sentiment. This research extends information completeness theory to visual data, highlighting that completeness emerges from functional image composition rather than quantity or aesthetics, offering new insights for multimodal persuasion and e-commerce design. Full article

Wangiri fraud is a pervasive telecommunications scam that exploits missed calls to lure victims into dialing premium-rate numbers, resulting in significant financial losses for operators and consumers. This paper presents a comprehensive machine learning framework for detecting Wangiri fraud in highly imbalanced and unlabeled Call Detail Record (CDR) datasets. We introduce a novel unsupervised labeling approach using domain-driven heuristics, coupled with advanced feature engineering to capture temporal, geographic, and behavioral patterns indicative of fraud. To address severe class imbalance, we evaluate multiple sampling strategies like the Synthetic Minority Over-sampling Technique (SMOTE) and undersampling, and also compare the performance of Logistic Regression, Decision Trees, Random Forest, XGBoost, and Multi-Layer Perceptron (MLP). Our results demonstrate that ensemble methods, particularly Random Forest and XGBoost, achieve near-perfect accuracy (e.g., Receiver Operating Characteristic Area Under the Curve (ROC-AUC) $>0.99$) on balanced data while maintaining interpretability. The proposed pipeline offers a scalable and practical solution for real-time fraud detection, providing telecom operators with an effective tool to mitigate Wangiri fraud risks. Full article

Background: Chest radiography is the most widely used diagnostic imaging modality globally, yet its interpretation is hindered by a critical shortage of radiologists, especially in low- and middle-income countries (LMICs). The interpretation is both time-consuming and error-prone in high-volume settings. Artificial Intelligence (AI) systems trained on public data may lack generalizability to multi-view, real-world, local images. Deep learning tools have the potential to augment radiologists by providing real-time decision support by overcoming these. Objective: We evaluated the diagnostic accuracy of a deep learning-based convolutional neural network (CNN) trained on multi-view, hybrid (public and local datasets) for detecting thoracic abnormalities in chest radiographs of adults presenting to a tertiary hospital, operating in offline mode. Methodology: A CNN was pretrained on public datasets (Vin Big, NIH) and fine-tuned on a local dataset from a Nepalese tertiary hospital, comprising frontal (PA/AP) and lateral views from emergency, ICU, and outpatient settings. The dataset was annotated by three radiologists for 14 pathologies. Data augmentation simulated poor-quality images and artifacts. Performance was evaluated on a held-out test set (N = 522) against radiologists’ consensus, measuring AUC, sensitivity, specificity, mean average precision (mAP), and reporting time. Deployment feasibility was tested via PACS integration and standalone offline mode. Results: The CNN achieved an overall AUC of 0.86 across 14 abnormalities, with 68% sensitivity, 99% specificity, and 0.93 mAP. Colored bounding boxes improved clarity when multiple pathologies co-occurred (e.g., cardiomegaly with effusion). The system performed effectively on PA, AP, and lateral views, including poor-quality ER/ICU images. Deployment testing confirmed seamless PACS integration and offline functionality. Conclusions: The CNN trained on adult CXRs performed reliably in detecting key thoracic findings across varied clinical settings. Its robustness to image quality, integration of multiple views and visualization capabilities suggest it could serve as a useful aid for triage and diagnosis. Full article

Accurate assessment of water quality is crucial for protecting public health and promoting environmental sustainability. Conventional laboratory-based methods for evaluating microbial contaminants are often time-consuming, resource-intensive, and reactive in nature, limiting their effectiveness for real-time water quality monitoring and management. This study examines the application of data-driven machine learning models to predict E. coli concentrations in Midmar Dam, utilizing readily available physicochemical parameters. A comparative analysis was conducted using five classical standalone ML algorithms: Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbors (kNN), Artificial Neural Network (ANN), and Extreme Gradient Boosting (XGBoost). These models were assessed based on their predictive performance using standard error metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). Among the models evaluated, the kNN algorithm demonstrated superior performance, achieving the lowest MSE and RMSE values, thereby highlighting its effectiveness in capturing the complex relationships between physicochemical indicators and microbial contamination levels. The findings demonstrate the potential of ML-based approaches to serve as efficient, scalable, and proactive tools for sustainable water-quality monitoring and management in dams. Full article

Unconfined Compressive Strength (UCS) of stabilized soils is commonly used for evaluating the effectiveness of soil improvement techniques. Achieving target UCS values through conventional trial-and-error approaches requires extensive laboratory experiments, which are time-consuming and resource-intensive. Automated Machine Learning (AutoML) frameworks offer a promising alternative by enabling automated, reproducible, and accessible predictive modeling of UCS values from more readily obtainable index and physical soil and stabilizer properties, reducing the reliance on experimental testing and empirical relationships, and allowing systematic exploration of multiple models and configurations. This study evaluates the predictive performance of five state-of-the-art AutoML frameworks (i.e., AutoGluon, AutoKeras, FLAML, H2O, and TPOT) using analyses of results from 10 experimental datasets comprising 2083 samples from laboratory experiments spanning diverse soil types, stabilizers, and experimental conditions across many countries worldwide. Comparative analyses revealed that FLAML achieved the highest overall performance (average PI score of 0.7848), whereas AutoKeras exhibited lower accuracy on complex datasets; AutoGluon , H2O and TPOT also demonstrated strong predictive capabilities, with performance varying with dataset characteristics. Despite the promising potential of AutoML, prior research has shown that fully automated frameworks have limited applicability to UCS prediction, highlighting a gap in end-to-end pipeline automation. The findings provide practical guidance for selecting AutoML tools based on dataset characteristics and research objectives, and suggest avenues for future studies, including expanding the range of AutoML frameworks and integrating interpretability techniques, such as feature importance analysis, to deepen understanding of soil–stabilizer interactions. Overall, the results indicate that AutoML frameworks can effectively accelerate UCS prediction, reduce laboratory workload, and support data-driven decision-making in geotechnical engineering. Full article

Background/Objectives: Delirium, an acute cognitive disturbance, is often unrecognized by family or friend care partners, contributing to delayed interventions and negative health outcomes. UnderstandingDelirium.ca is an e-learning lesson developed to address this gap by improving delirium knowledge among the public, patients, and family/friend care partners. Our objective was to evaluate the acceptability, intention to use, and perceived impact of Understanding Delirium e-learning among public users. Methods: A convergent mixed-methods observational evaluation combining survey-based quantitative data and thematic analysis was conducted. The survey included the Net Promoter Score (NPS), the short-form Information Assessment Method for patients and consumers (IAM4all-SF), and an open-text feedback item. Descriptive statistics were used to summarize IAM4all-SF responses, assessing perceived relevance, understandability, intended use, and anticipated benefit. Open-text comments were analyzed thematically by two independent reviewers who reached consensus through discussion. Subgroup analysis of qualitative themes was performed by age, gender, and NPS category. Results: Among 629 survey respondents, over 90% of respondents agreed that the lesson was relevant, understandable, likely to be used, and beneficial. The NPS was rated ‘excellent’ (score of 71), and lesson uptake included over 7000 unique users with a 35% completion rate. Qualitative analysis revealed themes of high educational value, emotional resonance, and perceived gaps in prior healthcare communication. Respondents emphasized the lesson’s clarity, intent to share, and potential for wider dissemination. Conclusions: UnderstandingDelirium.ca is a promising, guideline-aligned digital intervention that has potential to enhance delirium literacy and reduce care partner distress. Findings suggest that the Understanding Delirium e-learning can effectively improve public delirium literacy and should be integrated into care partner and clinical workflows. Full article

Background/Objectives: Breast cancer diagnosis based on histopathological examination remains the most reliable and widely accepted approach in clinical practice, despite being time-consuming and prone to inter-observer variability. While deep learning methods have achieved high accuracy in medical image classification, their cross-site generalization remains limited due to differences in staining protocols and image acquisition. This study aims to evaluate and compare three clinically relevant adaptation strategies to improve model robustness under domain shift. Methods: The ResNet50V2 model, pretrained on ImageNet and further fine-tuned on the Kaggle Breast Histopathology Images dataset, was subsequently adapted to the BreaKHis dataset under three clinically relevant transfer strategies: (i) threshold calibration without retraining (site calibration), (ii) head-only fine-tuning (light FT), and (iii) full fine-tuning (full FT). Experiments were performed on an internal balanced dataset and on the public BreaKHis dataset using strict patient-level splitting to avoid data leakage. Evaluation metrics included accuracy, precision, recall, F1-score, ROC-AUC, and PR-AUC, computed per magnification level (40×, 100×, 200×, 400×). Results: Full fine-tuning consistently yielded the highest performance across all magnifications, reaching up to 0.983 ROC-AUC and 0.980 sensitivity at 400×. At 40× and 100×, the model correctly identified over 90% of malignant cases, with ROC-AUC values of 0.9500 and 0.9332, respectively. Head-only fine-tuning led to moderate gains (e.g., sensitivity up to 0.859 at 200×), while threshold calibration showed limited improvements (ROC-AUC ranging between 0.60–0.73). Grad-CAM analysis revealed more stable and focused attention maps after full fine-tuning, though they did not always align with diagnostically relevant regions. Conclusions: Our findings confirm that full fine-tuning is essential for robust cross-site deployment of histopathology AI systems, particularly at high magnifications. Lighter strategies such as threshold calibration or head-only fine-tuning may serve as practical alternatives in resource-constrained environments where retraining is not feasible. Full article

This study introduces a machine learning-based framework for the automated determination of band gap energies in high-entropy oxides (HEOs) using UV-Vis spectroscopy data. Traditionally, band gap energies are obtained from the Tauc plots by manually extrapolating the linear region to the photon energy axis, a process that is time consuming and prone to human error, particularly when dealing with large datasets. To overcome these limitations, we developed a Python-based workflow that automates the band gap evaluation process through key steps, including data preprocessing, data augmentation, hyperparameter tuning, and band gap energy prediction. Various machine learning algorithms were employed to model the relationships between UV-Vis spectra and band gap energies, resulting in significant improvements in both accuracy and efficiency. Among the tested models, Bagging, Extra Trees, and Random Forest exhibited the best predictive performance, achieving mean absolute errors (MAE) as low as 0.26–0.28 eV and coefficients of determination (R²) of 0.73–0.74, substantially outperforming conventional automated methods. Although data augmentation and hyperparameter optimization yielded only modest performance gains, they contributed to improved model robustness. Overall, the proposed ML framework provides a scalable and efficient approach for the rapid characterization of HEOs, minimizing the need for manual analysis and accelerating data-driven materials discovery. Full article

Background/Objectives: Alzheimer’s disease (AD) accounts for ~70% of global dementia cases, with projections estimating 139 million affected individuals by 2050. This increasing burden highlights the urgent need for accessible, cost-effective diagnostic tools, particularly in low- and middle-income countries (LMICs). Traditional neuropsychological assessments, while effective, are resource-intensive and time-consuming. Methods: A total of 760 older adults (394 [51.8%] with AD) were recruited and neuropsychologically evaluated at the Instituto Colombiano de Neuropedagogía (ICN) in collaboration with Universidad del Norte (UN), Barranquilla. Machine learning (ML) algorithms were trained on a screening protocol incorporating demographic data and neuropsychological measures assessing memory, language, executive function, and praxis. Model performance was determined using 10-fold cross-validation. Variable importance analyses identified key predictors to develop optimized, abbreviated ML-based protocols. Metrics of compactness, cohesion, and separation further quantified diagnostic differentiation performance. Results: The eXtreme Gradient Boosting (xgbTree) algorithm achieved the highest diagnostic accuracy (91%) with the full protocol. Five ML-optimized screening protocols were also developed. The most efficient, the ICN-UN battery (including MMSE, Rey–Osterrieth Complex Figure recall, Rey Auditory Verbal Learning, Lawton & Brody Scale, and FAST), maintained strong diagnostic performance while reducing screening time from over four hours to under 25 min. Conclusions: The ML-optimized ICN-UN protocol offers a rapid, accurate, and scalable AD screening solution for LMICs. While promising for clinical adoption and earlier detection, further validation in diverse populations is recommended. Full article

Visible light positioning (VLP) has emerged as a promising indoor positioning technology, owing to its high accuracy and cost-effectiveness. In practical scenarios, signal attenuation, multiple light reflections, or light-deficient regions, particularly near room corners or furniture, can significantly degrade the light quality. In addition, the non-uniform light distribution by light-emitting diode (LED) luminaires can also introduce errors in VLP estimation. To mitigate these challenges, recent studies have increasingly explored the use of machine learning (ML) techniques to model the complex nonlinear characteristics of indoor optical channels and improve VLP performance. Convolutional neural networks (CNNs) have demonstrated strong potential in reducing positioning errors and improving system robustness under non-ideal lighting conditions. However, the performance of CNN-based systems is highly sensitive to their hyperparameters, including learning rate, dropout rate, batch size, and optimizer selection. Manual tuning of these parameters is not only time-consuming but also often suboptimal, particularly when models are applied to new or dynamic environments. Therefore, there is a growing need for automated optimization techniques that can adaptively determine optimal model configurations for VLP tasks. In this work, we propose and demonstrate a VLP system that integrates received signal strength (RSS) signal pre-processing, a CNN, and particle swarm optimization (PSO) for automated hyperparameter tuning. In the proof-of-concept VLP experiment, three different height layer planes (i.e., 200, 225, and 250 cm) are employed for the comparison of three different ML models, including linear regression (LR), an artificial neural network (ANN), and a CNN. For instance, the mean positioning error of a CNN + pre-processing model at the 200 cm receiver (Rx)-plane reduces from 9.83 cm to 5.72 cm. This represents an improvement of 41.81%. By employing a CNN + pre-processing + PSO, the mean error can be further reduced to 4.93 cm. These findings demonstrate that integrating PSO-based hyperparameter tuning with a CNN and RSS pre-processing significantly enhances positioning accuracy, reliability, and model robustness. This approach offers a scalable and effective solution for real-world indoor positioning applications in smart buildings and Internet of Things (IoT) environments. Full article