Heart rate variability in mental stress: The data reveal regression to the mean

Abstract

This data article aimed to assess whether there is a relationship between baseline heart rate variability (HRV) and mental stress-induced autonomic reactivity. Out of 1206 healthy subjects, 162 students were randomly selected to participate in this study. Participants were presented with a mental arithmetic task of 10 min duration. The task required serial subtraction of 7 from a randomly selected 3-digit number. During performance of this task as well as at baseline, ECG was recorded to acquire heart rate and HRV (high frequency, low frequency, the standard deviation of NN) data. Participants were divided into quartiles according to baseline HRV. Mental stress responses were compared across groups. We observed significant differences for autonomic reactivity scores between groups with high versus low baseline HRV. Linear regression results were consistent with the regression to the mean model and mental stress reaction (defined as mental stress value minus baseline value) negatively correlated with baseline values. Baseline-adjusted analyses did not demonstrate significant intergroup differences for changes in heart rate and HRV from rest to mental stress. These data suggest regression to the mean is a major source of variability of stress-related changes in heart rate variability.

Physiological correlates of cognitive load in laparoscopic surgery

“…The regression-to-the-mean which we observed in HRV (Fig. 6) has been shown in a mental arithmetic task 51 . To interpret this in another way, note that HRV has a lower bound near zero and, and HRV that is already low at baseline consequently has little room to further decrease under task conditions.…”

Quantitative Assessment of Stress Through EEG During a Virtual Reality Stress-Relax Session

“…Indeed, some of these works only report classification accuracy, overlooking intra-class performance, and they typically develop a single prediction model for all the participants, disregarding participant distinctiveness of brain activity. Alternatively, other works have addressed stress assessment through regression (Park et al, 2018;Ahuja and Banga, 2019;Dimitriev et al, 2019). Conversely to stress classification, these approaches do not aim to detect two or three stress levels, but to examine stress as a continuous variable.…”

“…Conversely to stress classification, these approaches do not aim to detect two or three stress levels, but to examine stress as a continuous variable. However, many of these works only assessed the linear relationship between certain biomarkers and self-perceived stress level (SPSL) (Saeed et al, 2017;Dimitriev et al, 2019), while others presented preliminary quantitative predictions of stress level that yielded middling results (for instance, in Park et al (2018), the authors obtained a correlation coefficient of 0.64, and in Das et al (2017), the authors did not provide a scoring metric for their regression predictions, although graphic representation of the predicted and actual stress values showed poor performance).…”

“…Additionally, we asked the participants to report their SPSL throughout the session according to an adaptation of the perceived stress scale (PSS) (Cohen et al, 1983). Finally, we used the SPSL surveys and the EEG biomarkers extracted from each participant to train and validate different regression algorithms, namely ridge regression (RR), random forest (RF), multi-layer perceptron (MLP), and support vector regression (SVR), that are usually applied in stress prediction literature (Khosrowabadi et al, 2011;Das et al, 2017;Ahn et al, 2019;Dimitriev et al, 2019).…”