Applying artificial neural network on heart rate variability and electroencephalogram signals to determine stress

Emotions are mental states, categorizes them into positive and negative feelings, and uses stress as an example of a negative emotion. Research demonstrated that acute and chronic stress can change physiological variables, such as heart rate variability (HRV) and electroencephalography (EEG). This research aims to early prevention and management of stress that is comfortable to use, reliable and accurate for stress detection. The Einthoven triangle rule was used to gather electrocardiogram (ECG) signals, while EEG signals were obtained from Fp1 and F3 connected to mikromedia 7 with the STM32F746ZG chipset. Various parameters were examined, including ECG signals in the time domain, frequency domain, non-linear analysis, and EEG signals in the frequency domain. Healthy subjects aged 18–23 undergoing different stress-inducing stages, with stress levels validated through the STAI-Y1 questionnaire. To process the HRV and EEG features, Pearson’s correlation function (PCF) was employed to select appropriated features into classification method. The proposed classification method in this research is the artificial neural network (ANN) with stratified K-fold, which yielded a stress level output accuracy of 95%. Additionally, the STAI-Y1 questionnaire results evaluation indicated a similarity score of 90.91%. This research has potential applications for individuals experiencing stress, providing a valuable tool for stress detection.