The Exploration of Predictors for Peruvian Teachers’ Life Satisfaction through an Ensemble of Feature Selection Methods and Machine Learning DOI Open Access
Luis Alberto Holgado-Apaza, Nelly Jacqueline Ulloa-Gallardo, Ruth-Nátaly Aragón-Navarrete

и другие.

Sustainability, Год журнала: 2024, Номер 16(17), С. 7532 - 7532

Опубликована: Авг. 30, 2024

Teacher life satisfaction is crucial for their well-being and the educational success of students, both essential elements sustainable development. This study identifies most relevant predictors among Peruvian teachers using machine learning. We analyzed data from National Survey Teachers Public Basic Education Institutions (ENDO-2020) conducted by Ministry Peru, filtering methods (mutual information, analysis variance, chi-square, Spearman’s correlation coefficient) along with embedded (Classification Regression Trees—CART; Random Forest; Gradient Boosting; XGBoost; LightGBM; CatBoost). Subsequently, we generated learning models Decision CatBoost; Support Vector Machine; Multilayer Perceptron. The results reveal that main are health, employment in an institution, living conditions can be provided family, performing teaching duties, as well age, degree confidence Local Management Unit (UGEL), participation continuous training programs, reflection on outcomes practice, work–life balance, number hours dedicated to lesson preparation administrative tasks. Among algorithms used, LightGBM Forest achieved best terms accuracy (0.68), precision (0.55), F1-Score Cohen’s kappa (0.42), Jaccard Score (0.41) LightGBM, (0.67), (0.54), (0.41), (0.41). These have important implications management public policy implementation. By identifying dissatisfied teachers, strategies developed improve and, consequently, quality education, contributing sustainability system. Algorithms such valuable tools management, enabling identification areas improvement optimizing decision-making.

Язык: Английский

COVID-19 health data prediction: a critical evaluation of CNN-based approaches DOI Creative Commons
Tae‐Hoon Kim,

Ravikumar Chinthaginjala,

Asadi Srinivasulu

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 17, 2025

The COVID-19 pandemic has significantly accelerated the demand for accurate and efficient prediction models to support effective disease management, containment strategies, informed decision-making. Predictive capable of analyzing complex health data are essential monitoring trends, evaluating risk factors, optimizing resource allocation during pandemic. Among various machine learning approaches, convolutional neural networks (CNNs) have emerged as powerful tools due their ability process large volumes high-dimensional data, such medical images, time-series patient demographics, with impressive precision. This research seeks systematically examine challenges limitations inherent in utilizing CNNs prediction, offering a comprehensive perspective grounded science research. Key areas investigation include issues related quality availability, incomplete, noisy, imbalanced datasets, which often hinder training robust models. Additionally, architectural constraints CNNs, including sensitivity hyperparameter tuning reliance on substantial computational resources, explored critical bottlenecks that impact scalability efficiency. A significant focus is placed generalization challenges, where trained specific datasets struggle adapt unseen from diverse populations or clinical settings, limiting applicability real-world scenarios. study further highlights reported accuracy 63%, underscoring need improved methodologies enhance model performance reliability. By addressing these this aims provide actionable insights practical recommendations optimize use prediction. In particular, emphasizes importance incorporating advanced strategies transfer learning, augmentation, regularization techniques overcome dataset robustness. integration multimodal approaches combining images auxiliary demographics laboratory results, proposed improve contextual understanding diagnostic Finally, underscores necessity interdisciplinary collaboration, leveraging domain expertise scientists, healthcare professionals, epidemiologists develop holistic solutions tackling complexities shedding light potential domain, guide researchers practitioners making decisions about design, implementation, optimization. Ultimately, it contributes advancing AI-driven diagnostics predictive modeling other public crises, fostering development scalable reliable better outcomes.

Язык: Английский

Процитировано

1
Boosting Skin Cancer Classification: A Multi-Scale Attention and Ensemble Approach with Vision Transformers DOI Creative Commons
Guang Yang, Suhuai Luo, Peter B. Greer

и другие.

Sensors, Год журнала: 2025, Номер 25(8), С. 2479 - 2479

Опубликована: Апрель 15, 2025

Skin cancer is a significant global health concern, with melanoma being the most dangerous form, responsible for majority of skin cancer-related deaths. Early detection critical, as it can drastically improve survival rates. While deep learning models have achieved impressive results in classification, there remain challenges accurately distinguishing between benign and malignant lesions. In this study, we introduce novel multi-scale attention-based performance booster inspired by Vision Transformer (ViT) architecture, which enhances accuracy both ViT convolutional neural network (CNN) models. By leveraging attention maps to identify discriminative regions within lesion images, our method improves models’ focus on diagnostically relevant areas. Additionally, employ ensemble techniques combine outputs several using voting. Our classifier, consisting EfficientNet models, classification 95.05% ISIC2018 dataset, outperforming individual The demonstrate effectiveness integrating methods classification.

Язык: Английский

Процитировано

0
SmartSkin-XAI: An Interpretable Deep Learning Approach for Enhanced Skin Cancer Diagnosis in Smart Healthcare DOI Creative Commons
Sultanul Arifeen Hamim,

Mubasshar U. I. Tamim,

M. F. Mridha

и другие.

Diagnostics, Год журнала: 2024, Номер 15(1), С. 64 - 64

Опубликована: Дек. 30, 2024

Background: Skin cancer, particularly melanoma, poses significant challenges due to the heterogeneity of skin images and demand for accurate interpretable diagnostic systems. Early detection effective management are crucial improving patient outcomes. Traditional AI models often struggle with balancing accuracy interpretability, which critical clinical adoption. Methods: The SmartSkin-XAI methodology incorporates a fine-tuned DenseNet121 model combined XAI techniques interpret predictions. This approach improves early by offering transparent decision-making process. was evaluated using two datasets: ISIC dataset Kaggle dataset. Performance metrics such as classification accuracy, precision, recall, F1 score were compared against benchmark models, including DenseNet121, InceptionV3, esNet50. Results: achieved 97% on 98% demonstrated high stability in measures, outperforming models. These results underscore robustness applicability real-world healthcare scenarios. Conclusions: addresses melanoma diagnosis integrating state-of-the-art architecture methods, providing both interpretability. enhances decision-making, fosters trust among professionals, represents advancement incorporating AI-driven diagnostics into medicine, bedside applications.

Язык: Английский

Процитировано

2
The Exploration of Predictors for Peruvian Teachers’ Life Satisfaction through an Ensemble of Feature Selection Methods and Machine Learning DOI Open Access
Luis Alberto Holgado-Apaza, Nelly Jacqueline Ulloa-Gallardo, Ruth-Nátaly Aragón-Navarrete

и другие.

Sustainability, Год журнала: 2024, Номер 16(17), С. 7532 - 7532

Опубликована: Авг. 30, 2024

Teacher life satisfaction is crucial for their well-being and the educational success of students, both essential elements sustainable development. This study identifies most relevant predictors among Peruvian teachers using machine learning. We analyzed data from National Survey Teachers Public Basic Education Institutions (ENDO-2020) conducted by Ministry Peru, filtering methods (mutual information, analysis variance, chi-square, Spearman’s correlation coefficient) along with embedded (Classification Regression Trees—CART; Random Forest; Gradient Boosting; XGBoost; LightGBM; CatBoost). Subsequently, we generated learning models Decision CatBoost; Support Vector Machine; Multilayer Perceptron. The results reveal that main are health, employment in an institution, living conditions can be provided family, performing teaching duties, as well age, degree confidence Local Management Unit (UGEL), participation continuous training programs, reflection on outcomes practice, work–life balance, number hours dedicated to lesson preparation administrative tasks. Among algorithms used, LightGBM Forest achieved best terms accuracy (0.68), precision (0.55), F1-Score Cohen’s kappa (0.42), Jaccard Score (0.41) LightGBM, (0.67), (0.54), (0.41), (0.41). These have important implications management public policy implementation. By identifying dissatisfied teachers, strategies developed improve and, consequently, quality education, contributing sustainability system. Algorithms such valuable tools management, enabling identification areas improvement optimizing decision-making.

Язык: Английский

Процитировано

1