Driver drowsiness is one of the leading causes of traffic accidents. This paper proposes a new method for classifying driver drowsiness using deep convolution neural networks trained by wavelet scalogram images of electrocardiogram (ECG) signals. Three different classes were defined for drowsiness based on video observation of driving tests performed in a simulator for manual and automated modes. The Bayesian optimization method is employed to optimize the hyperparameters of the designed neural networks, such as the learning rate and the number of neurons in every layer. To assess the results of the deep network method, heart rate variability (HRV) data is derived from the ECG signals, some features are extracted from this data, and finally, random forest and k-nearest neighbors (KNN) classifiers are used as two traditional methods to classify the drowsiness levels. Results show that the trained deep network achieves balanced accuracies of about 77% and 79% in the manual and automated modes, respectively. However, the best obtained balanced accuracies using traditional methods are about 62% and 64%. We conclude that designed deep networks working with wavelet scalogram images of ECG signals significantly outperform KNN and random forest classifiers which are trained on HRV-based features.

Cardiorespiratory interactions (CRIs) reflect the mutual tuning of two important organismic oscillators—the heartbeat and respiration. These interactions can be used as a powerful tool to characterize the self-organizational and recreational quality of sleep. In this randomized, blinded and cross-over design study, we investigated CRIs in 15 subjects over a total of 253 nights who slept in beds made from different materials. One type of bed, used as control, was made of melamine faced chipboard with a wood-like appearance, while the other type was made of solid wood from stone pine (Pinus cembra). We observed a significant increase of vagal activity (measured by respiratory sinus arrhythmia), a decrease in the heart rate (as an indicator of energy consumption during sleep) and an improvement in CRIs, especially during the first hours of sleep in the stone pine beds as compared to the chipboard beds. Subjective assessments of study participants’ well-being in the morning and sub-scalar assessments of their intrapsychic stability were significantly better after they slept in the stone pine bed than after they slept in the chipboard bed. Our observations suggest that CRIs are sensitive to detectable differences in indoor settings that are relevant to human health. Our results are in agreement with those of other studies that have reported that exposure to volatile phytochemical ingredients of stone pine (α-pinene, limonene, bornyl acetate) lead to an improvement in vagal activity and studies that show a reduction in stress parameters upon contact with solid wood surfaces.

The development of automated driving is an ongoing process; nonetheless, certain problems remain unresolved. One of them is the question when the automated vehicle control system should hand over the control to a human driver and whether this ...

Driver drowsiness is one of the leading causes of traffic accidents. This paper proposes a new method for classifying driver drowsiness using deep convolution neural networks trained by wavelet scalogram images of electrocardiogram (ECG) signals. Three different classes were de-fined for drowsiness based on video observation of driving tests performed in a simulator for manual and automated modes. The Bayesian optimization method is employed to optimize the hyperparameters of the designed neural networks, such as the learning rate and the number of neurons in every layer. To assess the results of the deep network method, Heart Rate Variability (HRV) data is derived from the ECG signals, some features are extracted from this data, and finally, random forest and k-nearest neighbors (KNN) classifiers are used as two traditional methods to classify the drowsiness levels. Results show that the trained deep network achieves balanced accuracies of about 77% and 79% in the manual and automated modes, respectively. However, the best obtained balanced accuracies using traditional methods are about 62% and 64%. We conclude that designed deep networks working with wavelet scalogram images of ECG signals significantly outperform KNN and random forest classifiers which are trained on HRV-based features.