Driving Stress Detection Using Multimodal CNN with Nonlinear Representation of Short-Term Physiological Signals

The research paper focuses on the problem of detecting driving stress using physiological signals. Traditional methods of stress detection are unreliable and time-consuming, hence there is a need for an automated and objective method that can help improve road safety and driving experience. The paper aims to address the problem of detecting driving stress using physiological signals, including PPG, ECG and EEG.

References:
Toward soft real-time stress detection using wrist-worn devices for human workspaces
Article
Full-text available
Feb 2021SOFT COMPUT
Sunder Ali Khowaja
Feri Setiawan
Seok-Lyong Lee
Bernardo Nugroho Yahya
View
Show abstract
Influence of mental stress on the pulse wave features of photoplethysmograms
Article
Full-text available
Nov 2019
Patrick Celka
Peter H Charlton
Bushra Farukh
Jordi Alastruey

Effects of stress on the development and progression of cardiovascular disease, Dec 2017, Mika Kivimäki, Andrew Steptoe

A critical review of proactive detection of driver stress levels based on multimodal measurements. ACM Comput. Surv. 2018

1. DRIVING STRESS DETECTION USING
MULTIMODAL CNN WITH NONLINEAR
REPRESENTATION OF SHORT-TERM
PHYSIOLOGICAL SIGNALS
Guidedby:Dr.RojalinaPriyadarshini
ASSOC.PROF.&HOD
Presentedby:PranjalKumar
RegdNo:20010130
Branch:CSE(2020-24)

2. OUTLINE
1
PROBLEM
STATEMENT
LITERATURE
REVIEW
PROPOSED
SOLUTION WORKFLOW
RESULT &
ANALYSIS
INTRODUCTION 2
3
4
5
6 SUMMARY
REFERENCES7
8

3. Driving stress is a common problem that affects many individuals and
can have serious consequences.
INTRODUCTION
Existing stress detection methods based on physiological signals are
limited in their accuracy and practicality.
The use of physiological signals, such as PPG, ECG, and EEG, has shown
promising results in detecting stress.
Previous studies have mainly focused on unimodal analysis. In this paper
multimodal approach using CNN is used.

4. LITERATURE REVIEW
S.No Title
Effects of stress on the
development and
progression of
cardiovascular disease
A critical review of
proactive detection of
driver stress levels
based on multimodal
measurements.
Demerit
The paper is primarily a literature
review of previous studies on the topic
and does not include any original
research conducted by the authors.
It does not include a quantitative
comparison or evaluation of their
performance.
Proposed
It was a literature review
article and did not propose any
new techniques or algorithms
Multimodal data fusion
techniques that combine data
from multiple sources
Author
Kivimäki, M.; Steptoe
Rastgoo, M.N.; Nakisa,
B.; Rakotonirainy, A.;
Chandran, V.;
Tjondronegoro
Journal
Nat. Rev. Cardiol. 2018
ACM Comput. Surv.
2018
3
2
1
4
Driver stress level
detection using HRV
analysis
The study was conducted on a small
sample size of only 15 drivers, which
may not be representative of the
general population of drivers.
Support Vector Machine
(SVM) is used to classify the
stress levels of the driver
based on the selected
features
Munla, N.; Khalil, M.;
Shahin, A.; Mourad
International Conference
on Advances in Biomedical
Engineering, 2015
Continuous detection of
physiological stress with
commodity hardware
The paper relies on self-reported stress
levels, which may not always be
accurate and could affect the reliability
of the results.
The trained model is used to
continuously monitor the
wearer's stress levels in real-
time, with stress levels
updated every few seconds.
Mishra, V.; Pope, G.;
Lord, S.; Lewia, S.;
Lowens, B.; Caine, K.;
Sen, S.; Halter, R.; Kotz
ACM Trans. Comput.
Healthc. 2020

5. PROBLEM
STATEMENT
The research paper focuses on the problem of detecting driving
stress using multiple physiological signals. Traditional methods
of stress detection are unreliable and time-consuming, hence
there is a need for an automated and objective method that can
help improve road safety and driving experience.
The paper aims to address the problem of detecting driving
stress using physiological signals, including PPG and GSM.

6. PROPOSED
SOLUTION
Proposed solution combines GSR, and HR signals for stress detection in
driving scenarios
Nonlinear transformation of signals using RQA and kurtosis-based
filtering
Multimodal CNN architecture for joint learning of features with random
undersampling technique used for imbalanced data
Achieved high accuracy and outperformed other state-of-the-art
methods

7. WORKFLOW

8. The proposed multimodal achieved an average accuracy of 95.67% and
AUC of 0.987 in detecting driving stress for the 30-s input signals.
RESULT &
ANALYSIS
The fusion of HRV and PPG signals resulted in higher performance
compared to using only one modality.
The proposed method also outperformed the baseline SVM model, which
only used HRV features, with an accuracy improvement of 7.9%.
The study showed that the proposed method has potential for real-time
monitoring of driving stress, which could help improve road safety and
prevent accidents caused by stress-induced driver errors.

9. SUMMARY
The research paper focuses on the problem of detecting driving
stress using physiological signals. Traditional methods of stress
detection are unreliable and time-consuming, hence there is a
need for an automated and objective method that can help
improve road safety and driving experience.
The paper aims to address the problem of detecting driving
stress using physiological signals, including PPG, ECG and EEG.

10. REFERENCES
03:Kivimäki, M.; Steptoe, A. Effects of stress on the development and progression of cardio
vascular disease. Nat. Rev. Cardiol. 2018
02:Rastgoo, M.N.;. A critical review of proactive detection of driver stress levels based on
multimodal measurements. ACM Comput. Surv. 2018
01:Munla, N.; Khalil, M.; Shahin, A.; Mourad; Prabono, Driver stress level detection using
HRV analysis. ICABE. 2015
04:Mishra, V.; Pope, G.; Continuous detection of physiological stress with commodity hardware.
ACM Trans. Comput. Healthc. 2020

11. THANK YOU

12. VERIFICATION
View publication stats