Feature group partitioning: an approach for depression severity prediction with class balancing using machine learning algorithms

Artículo Materias > Ingeniería Universidad Europea del Atlántico > Investigación > Producción Científica
Universidad Internacional Iberoamericana México > Investigación > Producción Científica
Universidad Internacional Iberoamericana Puerto Rico > Investigación > Producción Científica
Universidad Internacional do Cuanza > Investigación > Artículos y libros
Universidad de La Romana > Investigación > Producción Científica Abierto Inglés In contemporary society, depression has emerged as a prominent mental disorder that exhibits exponential growth and exerts a substantial influence on premature mortality. Although numerous research applied machine learning methods to forecast signs of depression. Nevertheless, only a limited number of research have taken into account the severity level as a multiclass variable. Besides, maintaining the equality of data distribution among all the classes rarely happens in practical communities. So, the inevitable class imbalance for multiple variables is considered a substantial challenge in this domain. Furthermore, this research emphasizes the significance of addressing class imbalance issues in the context of multiple classes. We introduced a new approach Feature group partitioning (FGP) in the data preprocessing phase which effectively reduces the dimensionality of features to a minimum. This study utilized synthetic oversampling techniques, specifically Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic (ADASYN), for class balancing. The dataset used in this research was collected from university students by administering the Burn Depression Checklist (BDC). For methodological modifications, we implemented heterogeneous ensemble learning stacking, homogeneous ensemble bagging, and five distinct supervised machine learning algorithms. The issue of overfitting was mitigated by evaluating the accuracy of the training, validation, and testing datasets. To justify the effectiveness of the prediction models, balanced accuracy, sensitivity, specificity, precision, and f1-score indices are used. Overall, comprehensive analysis demonstrates the discrimination between the Conventional Depression Screening (CDS) and FGP approach. In summary, the results show that the stacking classifier for FGP with SMOTE approach yields the highest balanced accuracy, with a rate of 92.81%. The empirical evidence has demonstrated that the FGP approach, when combined with the SMOTE, able to produce better performance in predicting the severity of depression. Most importantly the optimization of the training time of the FGP approach for all of the classifiers is a significant achievement of this research. metadata Shaha, Tumpa Rani; Begum, Momotaz; Uddin, Jia; Yélamos Torres, Vanessa; Alemany Iturriaga, Josep; Ashraf, Imran y Samad, Md. Abdus mail SIN ESPECIFICAR, SIN ESPECIFICAR, SIN ESPECIFICAR, vanessa.yelamos@funiber.org, josep.alemany@uneatlantico.es, SIN ESPECIFICAR, SIN ESPECIFICAR (2024) Feature group partitioning: an approach for depression severity prediction with class balancing using machine learning algorithms. BMC Medical Research Methodology, 24 (1). ISSN 1471-2288

Texto
s12874-024-02249-8.pdf
Available under License Creative Commons Attribution.
Descargar (2MB)

URL Oficial: http://doi.org/10.1186/s12874-024-02249-8

Resumen

In contemporary society, depression has emerged as a prominent mental disorder that exhibits exponential growth and exerts a substantial influence on premature mortality. Although numerous research applied machine learning methods to forecast signs of depression. Nevertheless, only a limited number of research have taken into account the severity level as a multiclass variable. Besides, maintaining the equality of data distribution among all the classes rarely happens in practical communities. So, the inevitable class imbalance for multiple variables is considered a substantial challenge in this domain. Furthermore, this research emphasizes the significance of addressing class imbalance issues in the context of multiple classes. We introduced a new approach Feature group partitioning (FGP) in the data preprocessing phase which effectively reduces the dimensionality of features to a minimum. This study utilized synthetic oversampling techniques, specifically Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic (ADASYN), for class balancing. The dataset used in this research was collected from university students by administering the Burn Depression Checklist (BDC). For methodological modifications, we implemented heterogeneous ensemble learning stacking, homogeneous ensemble bagging, and five distinct supervised machine learning algorithms. The issue of overfitting was mitigated by evaluating the accuracy of the training, validation, and testing datasets. To justify the effectiveness of the prediction models, balanced accuracy, sensitivity, specificity, precision, and f1-score indices are used. Overall, comprehensive analysis demonstrates the discrimination between the Conventional Depression Screening (CDS) and FGP approach. In summary, the results show that the stacking classifier for FGP with SMOTE approach yields the highest balanced accuracy, with a rate of 92.81%. The empirical evidence has demonstrated that the FGP approach, when combined with the SMOTE, able to produce better performance in predicting the severity of depression. Most importantly the optimization of the training time of the FGP approach for all of the classifiers is a significant achievement of this research.

Tipo de Documento:	Artículo
Palabras Clave:	Machine learning; Depression prediction; Class balancing; Oversampling; SMOTE; ADASYN; Stratified cross validation; Burn depression checklist; Feature group partitioning
Clasificación temática:	Materias > Ingeniería
Divisiones:	Universidad Europea del Atlántico > Investigación > Producción Científica Universidad Internacional Iberoamericana México > Investigación > Producción Científica Universidad Internacional Iberoamericana Puerto Rico > Investigación > Producción Científica Universidad Internacional do Cuanza > Investigación > Artículos y libros Universidad de La Romana > Investigación > Producción Científica
Depositado:	17 Jun 2024 23:30
Ultima Modificación:	17 Jun 2024 23:30
URI:	https://repositorio.unic.co.ao/id/eprint/12751

Acciones (logins necesarios)

Ver Objeto

open

Ultra Wideband radar-based gait analysis for gender classification using artificial intelligence

Gender classification plays a vital role in various applications, particularly in security and healthcare. While several biometric methods such as facial recognition, voice analysis, activity monitoring, and gait recognition are commonly used, their accuracy and reliability often suffer due to challenges like body part occlusion, high computational costs, and recognition errors. This study investigates gender classification using gait data captured by Ultra-Wideband radar, offering a non-intrusive and occlusion-resilient alternative to traditional biometric methods. A dataset comprising 163 participants was collected, and the radar signals underwent preprocessing, including clutter suppression and peak detection, to isolate meaningful gait cycles. Spectral features extracted from these cycles were transformed using a novel integration of Feedforward Artificial Neural Networks and Random Forests , enhancing discriminative power. Among the models evaluated, the Random Forest classifier demonstrated superior performance, achieving 94.68% accuracy and a cross-validation score of 0.93. The study highlights the effectiveness of Ultra-wideband radar and the proposed transformation framework in advancing robust gender classification.

Producción Científica

Adil Ali Saleem mail , Hafeez Ur Rehman Siddiqui mail , Muhammad Amjad Raza mail , Sandra Dudley mail , Julio César Martínez Espinosa mail ulio.martinez@unini.edu.mx, Luis Alonso Dzul López mail luis.dzul@uneatlantico.es, Isabel de la Torre Díez mail ,

Saleem

open

Children's and adolescents' lifestyle factors associated with physical activity in five Mediterranean countries: the DELICIOUS project

Background: Physical activity in children and adolescents represents one of the most important lifestyle factors to determine current and future health. Aim: The aim of the study is to assess the lifestyle and dietary factors linked to physical activity in younger populations across five countries in the Mediterranean region. Design: A total of 2,011 parents of children and adolescents (age range 6–17 years) participating to a preliminary survey of the DELICIOUS project were investigated to determine children's adequate physical activity level (identified using the short form of the international physical activity questionnaire) as well as diet quality parameters [measured as Youth-Healthy Eating Index (Y-HEI)] and eating and lifestyle factors (i.e., meal habits, sleep duration, screen time, etc.). Logistic regression analyses were performed to assess the odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between variables of interest. Results: Younger children of younger parents currently working had higher rates and probability to have adequate physical activity. Multivariate analysis showed that children and adolescents who had breakfast (OR = 1.88, 95% CI: 1.38, 2.56) and often ate with their family (OR = 1.80, 95% CI: 0.90, 3.61) were more likely to have an adequate level of physical activity. Children and adolescents who reported a sleep duration (8–10 h) closest to the recommended one were significantly more likely to achieve adequate levels of physical activity (OR = 1.88, 95% CI: 1.38, 2.56). Conversely, those with more than 4 h of daily screen time were less likely to engage in adequate physical activity (OR = 0.77, 95% CI: 0.54, 1.10). Furthermore, children and adolescents in the highest tertile of YEHI scores showed a 60% greater likelihood of engaging in adequate physical activity (OR = 1.60, 95% CI: 1.27, 2.01). Conclusion: These results emphasize the importance of promoting healthy diet and lifestyle habits, including structured and high quality shared meals, sufficient sleep, and screen time moderation, as key strategies to support active behaviors in younger populations. Future interventions should focus on reinforcing these behaviors through parental guidance and community-based initiatives to foster lifelong healthy habits.

Producción Científica

Alice Rosi mail , Francesca Scazzina mail , Maria Antonieta Touriz Bonifaz mail , Francesca Giampieri mail francesca.giampieri@uneatlantico.es, Achraf Ammar mail , Khaled Trabelsi mail , Osama Abdelkarim mail , Mohamed Aly mail , Evelyn Frias-Toral mail , Juancho Pons mail , Laura Vázquez-Araújo mail , Josep Alemany Iturriaga mail josep.alemany@uneatlantico.es, Lorenzo Monasta mail , Nunzia Decembrino mail , Ana Mata mail , Adrián Chacón mail , Pablo Busó mail , Giuseppe Grosso mail ,

Rosi

open

Molecular mechanisms underlying the neuroprotective effects of polyphenols: implications for cognitive function

Polyphenols are naturally occurring compounds that can be found in plant-based foods, including fruits, vegetables, nuts, seeds, herbs, spices, and beverages, the use of which has been linked to enhanced brain health and cognitive function. These natural molecules are broadly classified into two main groups: flavonoids and non-flavonoid polyphenols, the latter including phenolic acids, stilbenes, and tannins. Flavonoids are primarily known for their potent antioxidant properties, which help neutralize harmful reactive oxygen species (ROS) in the brain, thereby reducing oxidative stress, a key contributor to neurodegenerative diseases. In addition to their antioxidant effects, flavonoids have been shown to modulate inflammation, enhance neuronal survival, and support neurogenesis, all of which are critical for maintaining cognitive function. Phenolic acids possess strong antioxidant properties and are believed to protect brain cells from oxidative damage. Neuroprotective effects of these molecules can also depend on their ability to modulate signaling pathways associated with inflammation and neuronal apoptosis. Among polyphenols, hydroxycinnamic acids such as caffeic acid have been shown to enhance blood-brain barrier permeability, which may increase the delivery of other protective compounds to the brain. Another compound of interest is represented by resveratrol, a stilbene extensively studied for its potential neuroprotective properties related to its ability to activate the sirtuin pathway, a molecular signaling pathway involved in cellular stress response and aging. Lignans, on the other hand, have shown promise in reducing neuroinflammation and oxidative stress, which could help slow the progression of neurodegenerative diseases and cognitive decline. Polyphenols belonging to different subclasses, such as flavonoids, phenolic acids, stilbenes, and lignans, exert neuroprotective effects by regulating microglial activation, suppressing pro-inflammatory cytokines, and mitigating oxidative stress. These compounds act through multiple signaling pathways, including NF-κB, MAPK, and Nrf2, and they may also influence genetic regulation of inflammation and immune responses at brain level. Despite their potential for brain health and cognitive function, polyphenols are often characterized by low bioavailability, something that deserves attention when considering their therapeutic potential. Future translational studies are needed to better understand the right dosage, the overall diet, the correct target population, as well as ideal formulations allowing to overcome bioavailability limitations.

Producción Científica

Justyna Godos mail , Giuseppe Carota mail , Giuseppe Caruso mail , Agnieszka Micek mail , Evelyn Frias-Toral mail , Francesca Giampieri mail francesca.giampieri@uneatlantico.es, Julién Brito Ballester mail julien.brito@uneatlantico.es, Maurizio Battino mail maurizio.battino@uneatlantico.es, Carmen Lilí Rodríguez Velasco mail carmen.rodriguez@uneatlantico.es, José L. Quiles mail jose.quiles@uneatlantico.es,

Godos

open

Enhanced FPGA-based smart power grid simulation using Heun and Piecewise analytic method

The increasing complexity of modern power systems requires engineers to design, build, and test equipment with a high degree of accuracy. The demand for precise equipment design, testing, and evaluation has reached extraordinary levels within modern power systems. To meet this challenge, engineers rely heavily on real-time simulators, which are essential tools for assessing power network dynamics. This study introduces a novel approach, an adaptable and cost-effective simulator, poised to revolutionize traditional hardware-in-the-loop (HIL) systems. Leveraging field-programmable gate arrays (FPGAs) and a comprehensive implementation of Heun and Piecewise analytic methods (PAM), provided simulator offers unparalleled capabilities for embedded real-time simulation of smart grids, ensuring swift and accurate measurements. Augmented by Python-based process simulation and integrated with industry-standard tools like Modelica and MATLAB, the proposed system promises versatility and efficiency. Through comprehensive testing, including rigorous evaluations of excitation system responses to diverse scenarios such as voltage set-point variations, automatic voltage regulator step responses, and fault conditions, we demonstrate the simulator’s robustness and precision. Experimental findings underscore its potential as an effective alternative to conventional HIL systems, marking a significant advancement in smart grid simulation technology.

Producción Científica

Urfa Gul mail , Hafiz Muhammad Raza Ur Rehman mail , Muhammad Junaid Gul mail , Gerardo Méndez Mezquita mail , Alina Eugenia Pascual Barrera mail alina.pascual@unini.edu.mx, Imran Ashraf mail ,

Gul

open

A systematic review of deep learning methods for community detection in social networks

Introduction: The rapid expansion of generated data through social networks has introduced significant challenges, which underscores the need for advanced methods to analyze and interpret these complex systems. Deep learning has emerged as an effective approach, offering robust capabilities to process large datasets, and uncover intricate relationships and patterns. Methods: In this systematic literature review, we explore research conducted over the past decade, focusing on the use of deep learning techniques for community detection in social networks. A total of 19 studies were carefully selected from reputable databases, including the ACM Library, Springer Link, Scopus, Science Direct, and IEEE Xplore. This review investigates the employed methodologies, evaluates their effectiveness, and discusses the challenges identified in these works. Results: Our review shows that models like graph neural networks (GNNs), autoencoders, and convolutional neural networks (CNNs) are some of the most commonly used approaches for community detection. It also examines the variety of social networks, datasets, evaluation metrics, and employed frameworks in these studies. Discussion: However, the analysis highlights several challenges, such as scalability, understanding how the models work (interpretability), and the need for solutions that can adapt to different types of networks. These issues stand out as important areas that need further attention and deeper research. This review provides meaningful insights for researchers working in social network analysis. It offers a detailed summary of recent developments, showcases the most impactful deep learning methods, and identifies key challenges that remain to be explored.

Producción Científica

Mohamed El-Moussaoui mail , Mohamed Hanine mail , Ali Kartit mail , Mónica Gracia Villar mail monica.gracia@uneatlantico.es, Helena Garay mail helena.garay@uneatlantico.es, Isabel de la Torre Díez mail ,

El-Moussaoui

Enlaces de interesse

Enlaces de interesses

Feature group partitioning: an approach for depression severity prediction with class balancing using machine learning algorithms

Resumen

Acciones (logins necesarios)

ÁREA DE CONHECIMENTO

ACESSO

Língua

Filtro