Detecting Early Signs of Drowsy Driving using an Artificial Neural Network on Electrocardiography Signals

Abstract

Objective: This study developed an artificial neural network (ANN) model that can detect an early sign of drowsy driving (fighting-off drowsiness) based on electrocardiography (ECG) signals.

Background: Detecting an early state of drowsy driving is very important to prevent vehicle accidents on the road by providing appropriate interventions to the driver.

Method: The ECG signals for forty-three participants (mean age: 23.1, SD: 1.6) were recorded while performing a simulator-based monotonous driving for 20 minutes, and the ECG for twenty participants (mean age: 23.2, SD: 1.3) who suffered drowsiness were used in further analysis. The three driver states (normal, fighting-off drowsiness, and drowsy) were determined through participant's subjective report and video recording analysis. Six ECG measures in time and frequency domains were derived from the ECG and pre-processed to compensate individual variations in heart response.

Results: The model was trained using a feedforward network with a scaled conjugate gradient, and its average accuracy was over 99% for the training and testing data.

Conclusion: This study showed that the ECG can be used as a biometric indicator for the detection of the driver's drowsiness condition.

Application: The proposed model would be useful to the development of drowsiness detection system that can provide early warning to the driver at the onset of drowsiness.