Full text

Turn on search term navigation

© 2025. This work is published under https://creativecommons.org/licenses/by/4.0/ (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.

Abstract

Often referred to as the internetof medical things (IoMT), this systemleverages a combination of biosensors,actuators, detectors, cloud-based andedge computing, machine intelligence,and communication networks to deliverreliable performance and enhance quality of life in smart societies. Keywords: Driving fatigue Extreme environment EEG signals Optimized XGBoost Convolutional neural network IoMT Introduction Accidental injuries will be the top cause of mortality and the eighth major cause of death worldwide in 2030.u According to estimates, driving fatigue is the leading cause of traffic accidents.3 Nowadays, drowsy driving fatalitiesare a global problem, since they pose a grave threat to people's lives and property. In recent years, the construction of a wireless physiological signal monitoring system with secure data exchange inside the health care system has been a significant and dynamic activity.12 The use of the internet of medical things (IoMT) architecture, specifically smart biosensors at the edge, to enable mobility and rapid access to people's data has had a substantial influence in recent years.13,14 Another difficulty is discovering the link between various disciplines and learning characteristics.14 During data collection, variations in EEG signals across individuals are attributed to sleep quality, brain activity, and undefined external interference sources. Yang et al developed a broad learning system based on complex networks, demonstrating high accuracy in distinguishing between alert and fatigued states.21 Du et al proposed a fuzzy convolutional neural network combining EEG and ECG, achieving high accuracy and stability in detecting fatigue under noisy conditions.22 Wang et al employed Partial Directed Coherence to extract graph features from EEG, resulting in an 87.16% accuracy in fatigue detection.23 Zhang et al introduced a graph convolutional neural network based on Partial Directed Coherence, automatically extracting topological brain network features and achieving a 96.01% accuracy in fatigue detection.24 Zeng et al explored epidermal electronic systems for non-invasive mental fatigue monitoring, attaining an 89% accuracy using machine learning algorithms.25 Jantan et al. presented a multi-model approach utilizing convolutional neural networks, achieving over 99% accuracy in detecting driver fatigue.26 Gao et al developed a network combining spatial, frequency, and temporal information with an attention mechanism, surpassing other models in accuracy27 Sangeetha et al. introduced a DL approach for detecting microsleep using various EEG signals, demonstrating high accuracy and real-world applicability28 Jingwei developed a fatigue detection system combining EEG and Electromyography signals,utilizing the ThinkGear ASIC Module and signal analysis for accurate fatigue detection and driver alerts.29 Wu et al presented a DL model employing a sparse autoencoder for pilot fatigue detection, achieving high accuracy in detecting mental fatigue.30 Zhang et al introduced an auto-weighting incremental random vector functional link network for driver fatigue detection, outperforming existing methods with incremental learning capabilities.31 Sedik et al. developed a fatigue detection system combining fast Fourier transform and discrete wavelet transform (DWT) for feature extraction and noise reduction, achieving high accuracy in fatigue detection.32 Abbas and Alsheddy analyzed and compared different IoT platforms for driver fatigue detection, providing insights and improvement suggestions.33 Liu et al proposed transfer learning algorithms across individuals for mental fatigue detection, achieving higher accuracy with reduced calibration needs.34 Ding et al. presented a ResNet3D DL model for driver fatigue detection using three prefrontal EEG channels, achieving a 79.45% accuracy35 Gao et al introduced a multi-dimensional feature fusion network for fatigue detection, achieving an 85.16% accuracy across various datasets.36 Wu et al employed AutoEncoder to extract features from EEG signals, compressing and representing EEG signals of pilots and using a SoftMax classifier for pilot fatigue detection, achieving a 91.68% accuracy.37 Wen et al utilized AutoEncoder for unsupervised feature learning of EEG signals and paired it with an AdaBoost classifier to detect fatigue from the DEAP dataset, achieving a 95.00% accuracy38 Ma et al. presented a method using Modified-PCANet for feature extraction and SVM for EEG signal classification, achieving a 95.14% accuracy on a self-collected EEG dataset.39 Rundo et al. used Stacked AutoEncoder for feature extraction and SoftMax for classification, achieving a 100% accuracy on a self-collected EEG dataset.40 Panwar et al employed GAN for generating synthetic data and SoftMax for classification, achieving a 67.00% accuracy on a self-collected EEG dataset within a Wasserstein GAN setup.41 Lee et al combined LSTM and CNN to extract temporal and spatial features from EEG signals, achieving an 86.00% accuracy in three-class EEG signal classification using SoftMax.42 The reviewed studies showcase diverse approaches to driver fatigue detection using EEG signals, with high accuracy achieved through hybrid methods, DL, and feature extraction techniques.

Details

Title
Smart IoT-driven biosensors for EEG-based driving fatigue detection: A CNN-XGBoost model enhancing healthcare quality
Author
Rezaee, Khosro 1 ; Nazerian, Asmar 2 ; Zadeh, Hossein Ghayoumi 3 ; Attar, Hani 4 ; Khosravi, Mohamadreza 5 ; Kanan, Mohammad

 Department ot Biomedical Engineering, Meybod University, Meybod, Iran 
 Department of Engineering, Islamic Azad University, Qods Branch, Tehran, Iran 
 Department of Electrical Engineering, Faculty of Engineering, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran 
 Faculty of Engineering, Zarqa University, Zarqa, Jordan 
 Department of Medical Physics and Engineering, School of Medicine, Shiraz University of Medical Sciences, Shiraz, Iran 
Pages
1-15
Publication year
2025
Publication date
2025
Publisher
Tabriz University of Medical Sciences
ISSN
22285652
e-ISSN
22285660
Source type
Scholarly Journal
Language of publication
English
ProQuest document ID
3199837151
Copyright
© 2025. This work is published under https://creativecommons.org/licenses/by/4.0/ (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.