Classification of the Stress Levels Based on Heartbeat Evoked Potentials – a Pilot Study

It is known that emotion affects the brain signals (e.g., EEG) and heart rate variability (HRV). However, few studies have explored how the heartbeat-evoked potentials (HEPs) may relate to stress, a type of emotion. Here we show that the features extracted from the HEPs can be used to classify the low and high stress levels induced using a mental arithmetic (MA) paradigm. The HEPs were derived from 22 channels of scalp EEGs and one channel of ECGs simultaneously recorded when the subjects were solving the simple or complex MA problems. The feature matrix was constructed from the HEP epochs differing significantly between the two types of MA problems. Among three machine learning models, the best average classification accuracy reaches 99.89%.