Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

4 Citations (Scopus)
117 Downloads (Pure)

Abstract

Driving is an activity that requires considerable alertness. Insufficient attention, imperfect perception, inadequate information processing, and sub-optimal arousal are possible causes of poor human performance. Understanding of these causes and the implementation of effective remedies is of key importance to increase traffic safety and improve driver's well-being. For this purpose, we used deep learning algorithms to detect arousal level, namely, under-aroused, normal and over-aroused for professional truck drivers in a simulated environment. The physiological signals are collected from 11 participants by wrist wearable devices. We presented a cost effective ground-truth generation scheme for arousal based on a subjective measure of sleepiness and score of stress stimuli. On this dataset, we evaluated a range of deep neural network models for representation learning as an alternative to handcrafted feature extraction. Our results show that a 7-layers convolutional neural network trained on raw physiological signals (such as heart rate, skin conductance and skin temperature) outperforms a baseline neural network and denoising autoencoder models with weighted F-score of 0.82 vs. 0.75 and Kappa of 0.64 vs. 0.53, respectively. The proposed convolutional model not only improves the overall results but also enhances the detection rate for every driver in the dataset as determined by leave-one-subject-out cross-validation.
Original languageEnglish
Title of host publication2017 IEEE International Conference on Data Mining Workshops (ICDMW)
PublisherIEEE
Pages486-493
Number of pages8
ISBN (Electronic)978-1-5386-3800-2
ISBN (Print)978-1-5386-3801-9
DOIs
Publication statusPublished - 18 Dec 2017
EventFifth Workshop on Data Mining in Biomedical Informatics and Healthcare (DMBIH 2017) - The Roosevelt New Orleans, New Orleans, United States
Duration: 18 Nov 201721 Nov 2017
Conference number: 5
https://www.oakland.edu/secs/dmbih-workshop-2017

Workshop

WorkshopFifth Workshop on Data Mining in Biomedical Informatics and Healthcare (DMBIH 2017)
Abbreviated titleDMBIH 2017
CountryUnited States
CityNew Orleans
Period18/11/1721/11/17
OtherIn conjunction with the IEEE ICDM 2017
Internet address

Fingerprint

Simulators
Skin
Truck drivers
Neural networks
Learning algorithms
Feature extraction
Wearable sensors
Costs
Temperature
Deep learning
Deep neural networks

Cite this

Saeed, A., Trajanovski, S., van Keulen, M., & van Erp, J. B. F. (2017). Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors. In 2017 IEEE International Conference on Data Mining Workshops (ICDMW) (pp. 486-493). IEEE. https://doi.org/10.1109/ICDMW.2017.69
Saeed, Aaqib ; Trajanovski, Stojan ; van Keulen, Maurice ; van Erp, Johannes Bernardus Fransiscus. / Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors. 2017 IEEE International Conference on Data Mining Workshops (ICDMW). IEEE, 2017. pp. 486-493
@inproceedings{24d4687fcb3f43e5af6ac9d0126b5db7,
title = "Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors",
abstract = "Driving is an activity that requires considerable alertness. Insufficient attention, imperfect perception, inadequate information processing, and sub-optimal arousal are possible causes of poor human performance. Understanding of these causes and the implementation of effective remedies is of key importance to increase traffic safety and improve driver's well-being. For this purpose, we used deep learning algorithms to detect arousal level, namely, under-aroused, normal and over-aroused for professional truck drivers in a simulated environment. The physiological signals are collected from 11 participants by wrist wearable devices. We presented a cost effective ground-truth generation scheme for arousal based on a subjective measure of sleepiness and score of stress stimuli. On this dataset, we evaluated a range of deep neural network models for representation learning as an alternative to handcrafted feature extraction. Our results show that a 7-layers convolutional neural network trained on raw physiological signals (such as heart rate, skin conductance and skin temperature) outperforms a baseline neural network and denoising autoencoder models with weighted F-score of 0.82 vs. 0.75 and Kappa of 0.64 vs. 0.53, respectively. The proposed convolutional model not only improves the overall results but also enhances the detection rate for every driver in the dataset as determined by leave-one-subject-out cross-validation.",
author = "Aaqib Saeed and Stojan Trajanovski and {van Keulen}, Maurice and {van Erp}, {Johannes Bernardus Fransiscus}",
year = "2017",
month = "12",
day = "18",
doi = "10.1109/ICDMW.2017.69",
language = "English",
isbn = "978-1-5386-3801-9",
pages = "486--493",
booktitle = "2017 IEEE International Conference on Data Mining Workshops (ICDMW)",
publisher = "IEEE",
address = "United States",

}

Saeed, A, Trajanovski, S, van Keulen, M & van Erp, JBF 2017, Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors. in 2017 IEEE International Conference on Data Mining Workshops (ICDMW). IEEE, pp. 486-493, Fifth Workshop on Data Mining in Biomedical Informatics and Healthcare (DMBIH 2017), New Orleans, United States, 18/11/17. https://doi.org/10.1109/ICDMW.2017.69

Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors. / Saeed, Aaqib; Trajanovski, Stojan; van Keulen, Maurice ; van Erp, Johannes Bernardus Fransiscus.

2017 IEEE International Conference on Data Mining Workshops (ICDMW). IEEE, 2017. p. 486-493.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors

AU - Saeed, Aaqib

AU - Trajanovski, Stojan

AU - van Keulen, Maurice

AU - van Erp, Johannes Bernardus Fransiscus

PY - 2017/12/18

Y1 - 2017/12/18

N2 - Driving is an activity that requires considerable alertness. Insufficient attention, imperfect perception, inadequate information processing, and sub-optimal arousal are possible causes of poor human performance. Understanding of these causes and the implementation of effective remedies is of key importance to increase traffic safety and improve driver's well-being. For this purpose, we used deep learning algorithms to detect arousal level, namely, under-aroused, normal and over-aroused for professional truck drivers in a simulated environment. The physiological signals are collected from 11 participants by wrist wearable devices. We presented a cost effective ground-truth generation scheme for arousal based on a subjective measure of sleepiness and score of stress stimuli. On this dataset, we evaluated a range of deep neural network models for representation learning as an alternative to handcrafted feature extraction. Our results show that a 7-layers convolutional neural network trained on raw physiological signals (such as heart rate, skin conductance and skin temperature) outperforms a baseline neural network and denoising autoencoder models with weighted F-score of 0.82 vs. 0.75 and Kappa of 0.64 vs. 0.53, respectively. The proposed convolutional model not only improves the overall results but also enhances the detection rate for every driver in the dataset as determined by leave-one-subject-out cross-validation.

AB - Driving is an activity that requires considerable alertness. Insufficient attention, imperfect perception, inadequate information processing, and sub-optimal arousal are possible causes of poor human performance. Understanding of these causes and the implementation of effective remedies is of key importance to increase traffic safety and improve driver's well-being. For this purpose, we used deep learning algorithms to detect arousal level, namely, under-aroused, normal and over-aroused for professional truck drivers in a simulated environment. The physiological signals are collected from 11 participants by wrist wearable devices. We presented a cost effective ground-truth generation scheme for arousal based on a subjective measure of sleepiness and score of stress stimuli. On this dataset, we evaluated a range of deep neural network models for representation learning as an alternative to handcrafted feature extraction. Our results show that a 7-layers convolutional neural network trained on raw physiological signals (such as heart rate, skin conductance and skin temperature) outperforms a baseline neural network and denoising autoencoder models with weighted F-score of 0.82 vs. 0.75 and Kappa of 0.64 vs. 0.53, respectively. The proposed convolutional model not only improves the overall results but also enhances the detection rate for every driver in the dataset as determined by leave-one-subject-out cross-validation.

U2 - 10.1109/ICDMW.2017.69

DO - 10.1109/ICDMW.2017.69

M3 - Conference contribution

SN - 978-1-5386-3801-9

SP - 486

EP - 493

BT - 2017 IEEE International Conference on Data Mining Workshops (ICDMW)

PB - IEEE

ER -

Saeed A, Trajanovski S, van Keulen M, van Erp JBF. Deep Physiological Arousal Detection in a Driving Simulator Using Wearable Sensors. In 2017 IEEE International Conference on Data Mining Workshops (ICDMW). IEEE. 2017. p. 486-493 https://doi.org/10.1109/ICDMW.2017.69