Classifying Driver Distraction with Textile Electrocardiograms.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Human-Computer Interaction

Background:

• Continuous vital sign monitoring is crucial for driver safety.
• Textile sensors offer unobtrusive and non-invasive physiological data acquisition.
• Assessing driver mental wellbeing and distraction is essential for preventing accidents.

Purpose of the Study:

• To evaluate the effectiveness of a non-medical-grade textile electrocardiography (ECG) shirt for detecting driver distraction induced by stress.
• To investigate the utility of single-lead ECG signals from wearable sensors in real-time driver monitoring.
• To determine if ECG data can differentiate between baseline driving and distracted driving states.

Main Methods:

• Acquired single-lead ECG data from 10 healthy volunteers using ECG shirts in a controlled driving environment.
• Simulated three driving conditions: baseline, texting, and calling.
• Processed segmented ECG data (10, 30, 60 seconds) using a customized convolution neural network (ccNN).

Main Results:

• The ccNN model achieved a weighted F-Score of 0.65 and an average accuracy of 67.12% on the validation set.
• Leave-One-Subject-Out Cross-Validation demonstrated weighted F-Scores ranging from 0.53 to 0.75.
• The study indicates that wearable textile ECG signals contain informative patterns related to driver distraction.

Conclusions:

• A single-lead wearable textile ECG system can effectively provide insights into a driver's mental state and detect distraction.
• This technology holds potential for enhancing driver safety systems through continuous, non-invasive monitoring.
• Further research can explore advanced signal processing and machine learning techniques for improved accuracy in real-world driving scenarios.