Prediction of Perceived Stress Scores Using Low-Channel Electroencephalography Headband

Stress is a growing concern in this age of technological progress and productivity. Stress is traditionally measured using questionnaires with the help of psychologists. Due to shortage of health facilities and stigma associated with mental checkup, stress measurement process needs to be automated. This paper presents the results of an experiment to predict the perceived stress scores from electroencephalography (EEG) using a linear regression model. EEG signals are recorded using a five channel EEG headband for a duration of three minutes, with dry electrodes. In the first step, we acquire data by the Perceived Stress Scale (PSS) questionnaire and in the second step, we record EEG using muse headband and apply linear regression algorithm. The values of the PSS score serve as ground truth, which is then predicted by linear regression model. From the results, it is shown that the value of root mean squared error is as low as 2.36. This shows that PSS scores can be predicted with considerable accuracy by applying linear regression model on EEG signals. Our research suggests that perceived stress can be estimated from EEG signals using a low-channel headset once the system is trained.