TYPE Original Research PUBLISHED 10 May 2023 DOI 10.3389/fphys.2023.1147260 Assessing effectiveness of heart OPEN ACCESS rate variability biofeedback to EDITED BY Rui Min, Beijing Normal University, China mitigate mental health symptoms: REVIEWED BY Richard Gevirtz, a pilot study Alliant International University, United States Alireza Mani, Thais Castro Ribeiro 1,2*, Pau Sobregrau Sangrà 3, University College London, Esther García Pagès 2, Llorenç Badiella 4, Beatriz López-Barbeito 5, United Kingdom Sira Aguiló 5 and Jordi Aguiló 1,2 *CORRESPONDENCE Thais Castro Ribeiro, 1 Biomedical Research Network Center in Biogineering, Biomaterial and Nanomedicine (CIBER-BBN), thais.castro@uab.cat Madrid, Spain, 2Department of Microelectronics and Electronic Systems, Autonomous University of Barcelona, Barcelona, Spain, 3Clínic Foundation for Biomedical Research, Hospital Clínic of Barcelona, RECEIVED 18 January 2023 Barcelona, Spain, 4Applied Statistics Service, Autonomous University of Barcelona, Barcelona, Spain, ACCEPTED 24 April 2023 5 Emergency Department, Hospital Clínic of Barcelona, Barcelona, Spain PUBLISHED 10 May 2023 CITATION Castro Ribeiro T, Sobregrau Sangrà P, García Pagès E, Badiella L, López-Barbeito B, Aguiló S and Aguiló J Introduction: The increasing burden on mental health has become a worldwide (2023), Assessing effectiveness of heart concern especially due to its substantial negative social and economic impact. The rate variability biofeedback to mitigate mental health symptoms: a pilot study. implementation of prevention actions and psychological interventions is crucial to Front. Physiol. 14:1147260. mitigate these consequences, and evidence supporting its effectiveness would doi: 10.3389/fphys.2023.1147260 facilitate a more assertive response. Heart rate variability biofeedback (HRV-BF) COPYRIGHT has been proposed as a potential intervention to improve mental wellbeing © 2023 Castro Ribeiro, Sobregrau Sangrà, through mechanisms in autonomic functioning. The aim of this study is to García Pagès, Badiella, López-Barbeito, Aguiló and Aguiló. This is an open-access propose and evaluate the validity of an objective procedure to assess the article distributed under the terms of the effectiveness of a HRV-BF protocol in mitigating mental health symptoms in a Creative Commons Attribution License sample of frontline HCWs (healthcare workers) who worked in the COVID-19 (CC BY). The use, distribution or reproduction in other forums is pandemic. permitted, provided the original author(s) and the copyright owner(s) are credited Methods: A prospective experimental study applying a HRV-BF protocol was and that the original publication in this conducted with 21 frontline healthcare workers in 5 weekly sessions. For journal is cited, in accordance with PRE–POST intervention comparisons, two different approaches were used to accepted academic practice. No use, distribution or reproduction is permitted evaluate mental health status: applying (a) gold-standard psychometric which does not comply with these terms. questionnaires and (b) electrophysiological multiparametric models for chronic and acute stress assessment. Results: After HRV-BF intervention, psychometric questionnaires showed a reduction in mental health symptoms and stress perception. The electrophysiological multiparametric also showed a reduction in chronic stress levels, while the acute stress levels were similar in PRE and POST conditions. A significant reduction in respiratory rate and an increase in some heart rate variability parameters, such as SDNN, LFn, and LF/HF ratio, were also observed after intervention. Conclusion: Our findings suggest that a 5-session HRV-BF protocol is an effective intervention for reducing stress and other mental health symptoms among frontline HCWs who worked during the COVID-19 pandemic. The electrophysiological multiparametric models provide relevant information about the current mental health state, being useful for objectively evaluating the effectiveness of stress-reducing interventions. Further research could Frontiers in Physiology 01 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 replicate the proposed procedure to confirm its feasibility for different samples and specific interventions. KEYWORDS stress objective assessment, effectiveness of interventions, electrophysiological model, heart rate variability biofeedback, healthcare workers, mental health, stress 1 Introduction et al., 2020); b) HRV parameters to assess ANS functioning (Heiss et al., 2021); and c) electrodermal activity (EDA) indexes to evaluate In recent years, mental health has become a global concern. The sympathetic activation (Scavone et al., 2020). We hypothesize that increasing prevalence of mental health disorders also impairs use of subjective measures and single-objective indices limits the physical health, causing disability and negative economic and understanding of the stress response; hence, a multimodal approach social impacts (Flores et al., 2018). The World Health is being proposed in the current study. Indices derived from Organization (WHO) recognizes that mental healthcare is an electrocardiogram (ECG), EDA, peripheral skin temperature and essential part of comprehensive care, thus supporting the need respiration collectively may provide a more comprehensive for actions to promote mental wellbeing and implement information on physiological behavior and stress response prevention interventions (World Health Organization, 2022). (Garzón-Rey, 2017; Arza et al., 2018; García Pagès et al., 2023). There is substantial literature regarding psychological Our research group has suggested combined models that include a interventions (Purtle et al., 2020), although there is scarce set of electrophysiological variables, which were constructed from evidence supporting their validity, and limited quantitative statistical models based on the usual clinical psychological outcomes have been considered (Ahmed et al., 2021; Cairns assessment (self-reported questionnaires), thus providing a more et al., 2021). robust estimation (Garzón-Rey et al., 2017; Aguiló Mir et al., 2021). To advance in the field of mental health, it is crucial to establish The heart rate variability biofeedback (HRV-BF) is a proactive reliable tools that allow to objectively assess the patient’s affectability intervention that allows easy multimodal signal recording and and the effectiveness of applied interventions. The measurement of quantitative measurements, besides being broadly applied for electrophysiological variables provides information about the mental health issues (Moss and Shaffer, 2017). One of the HRV- functioning of the autonomic nervous system (ANS) and permits BF mechanisms is based on the physiological relation between estimating the stress level non-invasively, being an extended cardiovascular and respiratory systems, which is mediated by the quantitative approach for this purpose (Ahmed et al., 2022). vagus nerve, also known as respiratory sinus arrhythmia (RSA), A sustained stress response generates an ANS imbalance as the where the heart rate accelerates during inhalation and decelerates body tends to hyperactivate the sympathetic nervous system (SNS) during exhalation. Breathing at a certain respiratory rate—in general and the hypothalamus–pituitary–adrenal axis (HPA), leading to between 4.5 and 6.5 cycles per minute (resonance frequency)— hormonal and physiological adaptations for a “fight-or-flight stimulates the cardiovascular system’s resonant properties, response.” The parasympathetic nervous system (PNS) typically producing larger oscillations of heart rate, leading to higher HRV facilitates the recovery of stress responses, giving flexibility to in individuals and other beneficial effects, such as better gas adapt to different daily situations, although under a chronic exchange (Yasuma and Ichiro Hayano, 2004). stress condition, this function is compromised (McCraty and The other mechanism of HRV-BF is the stimulation of the Shaffer, 2015). baroreflex, which is also vagally regulated and responsible for Cardiac function is strongly regulated by the brain through blood pressure (BP) homeostasis. Upon an increase in BP, the dynamic interactions between the SNS and PNS (Tonhajzerova baroreflex induces a rapid adjustment in heart rate, reducing the et al., 2012). Heart rate variability (HRV), which reflects the blood flow and, subsequently, lowering BP (Lehrer, 2013). Again, fluctuation in time intervals between adjacent heartbeats, and its when breathing at the resonance frequency, the RSA stimulates metrics have become widely used indices of these interactions, and the baroreflex, thus strengthening its sensitivity. Therefore, hence, of psychological wellbeing. Low HRV has been associated HRV-BF training stimulates parasympathetic activity through with a diminished regulation capability of the ANS, cardiovascular both RSA and baroreflex, thus enhancing autonomic functioning outcomes, impairment of the immune system, aging, and other (Lehrer, 2013). conditions. Moreover, high HRV has been linked with better The HRV-BF applications are diverse; importantly, it has been performance, health, and adaptability (Kim et al., 2018). considered an effective intervention to reduce stress and anxiety (Yu Nevertheless, evidence showed limitations when interpreting et al., 2018; Kennedy and Parker, 2019), to mitigate posttraumatic HRV parameters in a simplistic binary way (sympathetic and stress symptoms (Schuman and Killian, 2019), and to reduce parasympathetic) and highlighted the complexity of this system symptoms of depression (LinMei et al., 2019; Pizzoli et al., 2021). and influencing factors, such as age and respiratory rate (Billman, It was also proposed as a promising intervention to reduce 2013; Hernando et al., 2016; Hayano and Yuda, 2019). physiological distress in healthcare workers (HCWs) in the In the literature, the outcomes more frequently selected to COVID-19 pandemic (Aristizabal et al., 2020); however, there is evaluate the effect of mental health interventions are as follows: no evidence in this context yet. Seeking to fill this gap and a) psychometric questionnaires to subjectively estimate the anxiety considering that frontline HCWs constitute a vulnerable group, state, stress level, and severity of depression, among others (Lehrer especially with the emergence of the COVID-19 pandemic Frontiers in Physiology 02 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 FIGURE 1 Scheme of the 5-session experimental protocol. The colors identify the different steps in each of the sessions. Those marked in gray represent the biofeedback training. (Giorgi et al., 2020; Pappa et al., 2020; Sobregrau Sangrà et al., 2022), The inclusion criteria were to be a HCW directly involved with the current study applied an HRV-BF protocol in this population. care of COVID-19 patients during the pandemic. The selection was The main aim of this study is to propose and validate an objective made according to a high risk of developing post-traumatic stress procedure to easily assess the effectiveness of mental health disorder or a moderate perception of stress and/or anxiety according interventions, specifically in this study of a personalized HRV-BF to cutoff points of current gold-standard questionnaires (i.e., a score protocol in mitigating mental health symptoms in frontline HCWs of 30 or higher on PCL5; 14 or higher on PSS; and 10 or higher on who worked in the COVID-19 pandemic. For this purpose, we designed STAI-S) and voluntarily agreeing to participate. a PRE and POST experimental study based on two different approaches of psychological assessments. First, considering the gold-standard psychometric questionnaires and second, applying a recently 2.2 Experimental procedure developed electrophysiological multiparametric model that has been validated to assess chronic stress levels. As a secondary objective, we The HRV-BF protocol was designed based on (Lehrer et al., propose to investigate whether an acute stress model can distinguish the 2013) and comprises 5 weekly in-person sessions of 45–60 min, different stages of physiological response assessment (baseline, stress including specific instructions for daily practices. Each session was exposure, and recovery), and also the most relevant variables for this held individually in a quiet room conducted by a trained purpose in this population. professional and scheduled according to participant availability. If a participant could not attend one of the sessions, it was rescheduled for as soon as possible in order to complete the protocol and to 2 Methods prevent a longer interval between sessions from compromising adherence to practice. Figure 1 shows the organization of the 2.1 Participants overall protocol by sessions. Participants provided information on demographics, working Twenty-one frontline HCWs were included in this study conditions, psychological background, use of medication, and social between March and July 2021, recruited from two Spanish habits. Self-reported psychometric questionnaires (yellow blocks tertiary hospitals: Hospital Clínic of Barcelona and Hospital del shown in Figure 1) were collected, and a physiological response Mar of Barcelona. The study was conducted at the Hospital Clínic of measurement (salmon-colored blocks) was carried out prior to Barcelona, and the protocol was approved by its Ethics Committee. (PRE) and immediately after (POST) the HRV-BF training to All participants gave a written informed consent. perform comparisons. Frontiers in Physiology 03 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 In the first session, preceding the training, the individual’s The reliability of these questionnaires was assessed for the study resonance frequency (RF), i.e., the optimal breathing rate, was sample by Cronbach’s alpha, and estimates were all above 0.85, thus determined based on the following criteria: phase synchrony showing a high degree of internal consistency (O’Rourke and between respiration and HR and signal smoothness, maximal Hatcher, 2013). HRV (higher peak–trough amplitude), greater power in low- frequency (LF) band, and a peak around 0.1 Hz on the LF band 2.3.2 Physiological variables (Lehrer et al., 2013; Shaffer and Meehan, 2020). RF was revised and Physiological monitoring provided an insight into differences in fine-tuned in the following sessions (lilac-colored blocks). autonomic functioning in a stress-inducing situation before and In sessions 2, 3, 4, and 5, the biofeedback training (gray blocks) after the intervention. The participants were asked to be seated in a was focused on specific breathing techniques, such as pursed lips and comfortable chair with armrests and eyes open. The physiological diaphragmatic breathing, guided by respiration and heart rate response measurement was analyzed in three different stages: 1) variability real-time waves displayed by using BioTrace + baseline measurement (BM), taken as a reference when the software (Mind Media, 2018 version). Participants were participant is in a resting state; 2) stress exposure (SE), to instructed to practice specific breathing exercises daily for evaluate the physiological reactivity facing stressors; and 3) 10–20 min (blue blocks). For home practice, the given recovery measurement (RM) to assess resilience after stressors. recommendations increased the difficulty level at each session: The SE stage included three different stressors: a) 2-min Stroop week 1: starting with brief periods of slow breathing practice task (originally Stroop, 1935), with an incongruent stimulus, in with a pacer to correctly follow the rhythm of the individual’s which the colors were written in an ink that does not match the resonance frequency. The pace of breathing would become more color-word, and the participant should inform the color of the ink; fluid and comfortable for the participant, and the practice time b) 2-min mental arithmetic task (Kirschbaum et al., 1993), which would be gradually increased; week 2: same as the previous one and consisted of performing mental calculations as quickly and adding diaphragmatic breathing awareness exercises, such as placing accurately as possible; in case of an error, the participant should one hand over the abdomen and the other over the chest and start over; c) 2-min telling about the experience of working on the breathing in such a way that the hand over the abdomen moves and frontlines of the COVID-19 pandemic and the feelings they could the other is as still as possible. It could be performed lying down to report (detailed in Figure 2). Differences between stressors were not facilitate the perception of the movement; week 3: performing the analyzed in this study. same exercise as before without using the hands and more frequently in a sitting position. One must start practicing using the pacer and 2.3.2.1 Data acquisition then performing brief periods of free practice; week 4: performing The electrophysiological data were recorded using the NeXus-10 breathing without using the pacer whenever they feel safe following MKII device (Mind Media BV, Herten, Netherlands) and the real- the already incorporated rhythm; then one must start applying it on time waves displayed by Biotrace+ software (Mind Media, day-to-day activities. In case of insecurity to do it freely, they should 2018 version) during each of the sessions. Electrocardiogram start the practice with the pacer and try to perform in short free (ECG), photoplethysmogram (PPG), electrodermal activity periods and gradually increase the duration; week 5: it is (EDA), skin temperature (ST), and respiration (Resp) were recommended to continue performing slow breathing on a daily signals simultaneously gathered. The electromyogram (EMG) of basis and applying it in everyday situations, especially after or before the upper trapezius muscle was monitored only for bringing a stressful event. The importance of daily practice to achieve awareness of muscular activation during breathing, although this physiological and long-term benefits is emphasized to engage the signal was not included in the study analyses. participants. The use of free applications on their mobiles was For the EDA, disposable electrodes were affixed to the palmar encouraged to practice RF at home with a pacer (Paced Breathing surface of the middle phalanges of the second and fourth digits; for for Android and Awesome Breathing: Pacer Time for iOS). PPG, the clip sensor was placed on the middle finger, and on the fingertip of the fifth finger was attached the ST sensor. The electrode was placed in the non-dominant hand to avoid excessive movements 2.3 Study outcomes and possible large artifacts. For lead 1 of the ECG signal, electrodes were placed below the 2.3.1 Self-reported psychometric questionnaires right collarbone and below the left rib cage, whereas for lead 2, Self-reported psychometric questionnaires consisted of gold- electrodes were placed on the fifth intercostal space along the mid- standard psychological questionnaires in validated Spanish versions axillary line on the left and the other symmetrically on the right side. to examine stress, anxiety, and depression under PRE–POST The reference electrode was placed on the left collarbone. conditions. Stress was examined using the Perceived Stress Scale To monitor the Resp signal, an adjustable elastic band was (PSS) (Remor, 2006) and the Visual Analog Stress Scale (VASS) placed over the abdomen and over clothing. This sensor detects (Lesage et al., 2012). The Posttraumatic Stress Disorder Checklist for abdominal expansion/contraction; for instance, when inhaling, the DSM-5 (PCL5) was applied as a screening for post-traumatic stress device stretches, and during exhalation, the sensor returns to its disorder (PTSD) symptoms (Blevins et al., 2015). Anxiety symptoms neutral position. were evaluated using the State-Trait Anxiety Inventory (STAI) Simultaneously, the PPG, EDA, and ST signals were monitored (Spielberger et al., 1983), while to assess depression, the Patient through the Empatica E4 wearable wristband (Empatica Boston, Health Questionnaire-9 (PHQ9) was used (Diez-Quevedo et al., United States). These data will be explored in further analysis 2001). All instrument descriptions are given in Table 1. looking for a more easy and unobtrusive way to assess the Frontiers in Physiology 04 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 TABLE 1 Self-reported instruments used to assess the outcomes. Outcome Instrument Description Stress perception Visual Analog Stress Scale (VASS) Visual 100-point scale (0, not at all; 100, absolutely stressed) Perceived Stress Scale (PSS) 10-item Likert scale (0, never; 4, very often) Risk of PTSD Post-traumatic Stress Disorder Checklist for DSM-5 (PCL-5) 20-item Likert scale (0, not at all; 4, extremely) Current level of anxiety State Anxiety Inventory (STAI-S) 40-item Likert scale (1, not at all; 4, very much so). Each subscale includes 20 items Trait anxiety Trait Anxiety Inventory (STAI-T) Depression Patient Health Questionnaire-9 (PHQ9) 9-item Likert scale (0, not at all; 3, nearly every day) FIGURE 2 Schematic procedure for physiological response measurement PRE- and POST-intervention. effectiveness of interventions, considering the number of HRV-BF computation of the RR series. The algorithm searches for sudden sessions and the interval between them. changes in interbeat intervals, performs an interpolation where it has found non-normal beats, and based on this, proposes a 2.3.2.2 Data processing correction to them. Segments of up to three interpolated/ The different electrophysiological raw signals mentioned were corrected beats have been accepted and are assumed to be analyzed in a 1-min window using BioSig browser in MATLAB normal. Following this, a time-domain analysis was performed to software (Garzón-Rey et al., 2017), and several variables were extract the mean heart rate (HRV-HR) in beats per minute; the extracted from each one of them as described in Table 2. For standard deviation of differences between adjacent normal peak comparisons, the mean of all 1-min segments was computed as a intervals (HRV-SDNN); and the root mean square of successive single value per stage for each subject. differences between beat intervals (HRV-RMSSD). For the In this study, the PPG signal will be used in conjunction with frequency-domain analysis, the low- and high-frequency absolute that of ECG to extract pulse arrival time (PAT) parameters. To powers (HRV-LF and HRV-HF, respectively) were computed from extract HRV parameters, the ECG signal was chosen since it the spectral density of the HRV signal, calculated using Fourier provides more accurate measurements. transform. The HRV signal is obtained by subtracting from the From the ECG signal, the beat was detected through a discrete instantaneous HR signal, a low-pass-filtered HR signal (cutoff wavelet transform (Martínez et al., 2004). Given the robustness of frequency of 0.03 Hz), which mainly explains the changes in the algorithm and the visual inspection of the quality of the ECG mean HR. The HRV–LF/HF ratio was computed by the ratio signals, no type of filtering was applied. Afterward, the existence of between absolute power in low- and high-frequency bands, ectopic beats or false QRS detections was verified and fixed using the i.e., HRV-LF/HRV-HF. The HRV-LFn and HRV-HFn represent algorithm reported in Mateo and Laguna (2003) prior to the the low- and high-frequency power normalized, respectively. Frontiers in Physiology 05 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 TABLE 2 Description of extracted parameters from electrophysiological variables. Electrophysiological signal Sample Extracted Description frequency (Hz) parameters Electrocardiogram (ECG) 1024 HRV-HR, bpm Mean heart rate HRV-SDNN, s Standard deviation of normal beat intervals HRV-RMSSD, s Root mean square of successive differences between beat intervals HRV-sdsd, s Standard deviation of differences between adjacent R–R peaks intervals HRV-VLF, s-2 Absolute power of the very low-frequency band (0.003–0.04 Hz) HRV-LF, s-2 Absolute power of the low-frequency band (0.04–0.15 Hz) -2 HRV-HF, s Absolute power of the high-frequency band (0.15–0.4 Hz) HRV-LF/HF ratio Ratio of low-frequency to high-frequency power HRV-LFn, nu Relative power of the low-frequency band normalized HRV-HFn, nu Relative power of the high-frequency band normalized Electrocardiogram (ECG) and 1024 PAT, ms Mean pulse arrival time, the time between the beat detected by ECG and photoplethysmography (PPG) the pulse by PPG stdPAT, ms Standard deviation of pulse arrival time Respiration (Resp) 512 RR, Hz Respiratory rate Pk, Hz Respiratory rate peak in the power spectrum Electrodermal activity (EDA) 512 Tonic, µS Average value of the tonic component, i.e., slowly changing level of conductance of the skin, also known as skin conductance level (SCL) Phasic, µS Average value of the phasic component, i.e., fast-changing responses typically associated with short-term events, also known as skin conductance responses (SCR) aucPhasic, µS·s Area under the curve of the phasic component, related to SCR EDASymp, µS Electrodermal response in the power spectrum (0.045–0.25 Hz) Skin temperature (ST) 512 TFinger, °C Mean finger temperature TGradient, °C Mean gradient of finger temperature TPower, °C 2 Mean power of finger temperature The PPG signal was filtered using a low-pass FIR filter with a affects the signal quality. The missing values were replaced by cut-off frequency of 35 Hz (order 50) and, then, a high-pass FIR median imputation to perform the comparisons between stages. filter with a cut-off frequency of 0.3 Hz (order 5000). PPG artifacts The EDA signal was visually inspected to delete motion artifacts were suppressed through a Hjorth-parameter-based PPG artifact and linearly interpolated. The windows with interpolated segments detector described in Gil et al. (2009). Pulses were detected from the larger than 25% were discarded, and the signal was then resampled PPG signal on those time slots without artifacts by an algorithm at 4 Hz. Then, EDA was processed in two different ways. First, a based on Lazaro et al. (2014). Subsequently, the mean time time-domain analysis was conducted using a convex optimization difference between the R peak in the ECG signal and the point model, called cvxEDA (Greco et al., 2016), which decomposes the of 50% increase, corresponding to the pulse detected on the finger by signal as a linear combination of the tonic and phasic components the PPG signal, was considered the pulse arrival time (PAT), and its (i.e., the skin conductance level and response, respectively) and some standard deviation (stdPAT) was also calculated. noise (incorporating the error of the model, artifacts, and other The respiratory rate (RR) was estimated as the frequency to measurement errors). The average value of the tonic (Tonic) and which the maximum peak (Pk) of the power density spectrum phasic (Phasic) components and the area under the curve of the corresponds, estimated by means of fast Fourier transform phasic component (aucPhasic) were calculated. The second way was (Lázaro et al., 2014). When the Pk was greater than 65%, then a frequency-domain analysis of EDA, proposed to assess the RR was accepted as the respiratory rate. The respiration wave sympathetic tone through a parameter named EDASymp was filtered with an FIR passband filter with cutoff frequencies of (Posada-Quintero et al., 2016). In this analysis, the EDA signal 0.03 and 0.9 Hz. In 10 participants, the Resp parameters could not be was filtered with an FIR bandpass filter with 0.01 and 0.9 Hz cutoff calculated during the SE because they did not meet the condition frequencies, and the power of the electrodermal response in the band aforementioned, probably due to the speech during the task, which 0.045–0.25 Hz (EDASymp) was computed. This bandwidth has Frontiers in Physiology 06 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 been reported to correspond to sympathetic dynamics in non- TABLE 3 Clinical and sociodemographic characteristics of participants. exercise conditions. Characteristics Mean N (%) Finally, for the skin temperature signal, visual inspection showed (SD) that the recorded signals did not contain significant artifacts; therefore, no pre-processing was needed. Apart from the skin Age (years) 37.7 (±11.7) temperature average value (TFinger), the gradient of the Protocol completion time (days) 36 (±11.4) successive differences of TFinger every 10 s was calculated Daily practice accomplishment (%) 62.3 (±19.1) (TGradient), indicating the speed of temperature change. In relation to the amplitude, the power of the signal calculated as BMI (kg/m2) 24.2 (±3.8) the mean of the square of the signal (TPower) is used. Normal (18.5–24.9) 12 (57.2%) Overweight (≥25) 6 (28.6%) 2.3.3 Multiparametric chronic and acute stress Obesity (≥30) 3 (14.3%) models Physical activity For a more robust estimation of the stress level, we used two Rarely 5 (23.8%) different models a) to assess the chronic stress level (Aguiló Mir Occasional 11 (52.4%) et al., 2021), called electrophysiological signal-based stress model Regular 5 (23.81%) (ESBSm) and b) to assess the acute stress level, called ES3 (Garzón- Rey, 2017). These models were calculated from the variables Increased substance use extracted from the electrophysiological signals explained in the Yes 10 (52.4%) previous section. No 11 (47.6%) The chronic stress level model (ESBSm) was designed to Working hours per week (hours) 38.3 (±3.6) match up with a reference scale including psychometric questionnaires and biochemical variables, taken as a gold- Professional category standard indicator of stress response. ESBSm includes Clinicians 6 (28.6%) 11 standardized parameters extracted from ECG, Resp, and Nurses 13 (61.9%) EDA and a few quadratic terms. Some parameters are Auxiliary nurses 2 (9.5%) previously logarithmically transformed to avoid data skewness Working unit effects. In the present study, this model is applied as an COVID-19 wards 6 (28.6%) instrument to assess the effectiveness of the HRV-BF protocol Emergency service 13 (61.9%) since it allows detecting changes in the chronic stress level ICU 2 (9.5%) over time. The acute stress level model (ES3) also fits the results of Employment category psychometric questionnaires and biochemical biomarkers in a Fixed-term 14 (66.7%) sample subjected to an induced acute stress task. The Open-ended 7 (33.3%) ES3 includes four parameters extracted from ECG, EDA, and ST. Living with people at high risk of COVID-19 In the present study, ES3 is used to assess whether changes occur infection between the different stages (BM, SE, and RM). As with the previous Yes 7 (33.3%) model, the analysis was also performed comparing values in PRE No 14 (66.7%) and POST conditions. Psychological support requested Yes 3 (14.3%) No 18 (85.7%) 2.4 Statistical analysis Working leave due to COVID-19 A descriptive statistical analysis was performed to summarize Yes 11 (52.4%) the sample characteristics. Continuous variables are presented as No 10 (47.6%) mean and standard deviation (SD), and categorical variables are Increase in weekly working hours presented as frequency and percentages. Yes 9 (42.9%) For the psychometric questionnaires and chronic stress model No 12 (57.1%) (ESBSm) data, differences in PRE–POST intervention were evaluated with paired t-tests or Wilcoxon signed-rank tests, Current stress-related medication according to data distribution. Yes 4 (19%) For the electrophysiological variables, in order to compare No 17 (81%) conditions (PRE and POST) and stages (BM, SE, and RM), a Previous affective disorders mixed model (repeated measures analysis of variance, ANOVA) Yes 7 (33.3%) was used. The compliance of the application criteria was assessed No 14 (66.7%) with the analysis of model residuals. Skewed variables were analyzed using log-transformation. (Continued on following page) Frontiers in Physiology 07 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 TABLE 3 (Continued) Clinical and sociodemographic characteristics of From the bivariate analysis, the psychometric questionnaires participants. showed a reduction in all scores after intervention, with exception of Characteristics Mean N (%) the STAI-T, which did reduce even though without statistical (SD) significance (p = 0.147). Comparisons between PRE- and POST- intervention are shown in Table 4. Pharmacological treatment for stress As exploratory analyses, we examined whether PRE–POST changes Yes 16 (76.2%) are influenced by age, BMI, physical activity, and daily practice. There No 5 (23.8%) are no significant findings. However, there is an effect related to the Previous treatment with psychotropics initial condition, i.e., those with high initial scores are more likely to Yes 4 (19%) show improvement in self-reported questionnaires (VASS: p = 0.003; No 17 (81%) PSS: p = 0.016; PCL5: p = 0.036; PHQ9: p = 0.016; STAI-S: p < 0.001), Key: BMI, body mass index; ICU, intensive care unit; SD, standard deviation. apart from STAI-T (p = 0.081). 3.2.2 Chronic stress measurement For PRE–POST comparisons, the baseline measurement (BM) Post-hoc tests with Tukey’s correction were performed to assess was considered to mimic the conditions of the ESBSm chronic stress the time and stage effects on physiological variables and on the acute study. Figure 4 illustrates a reduction (77.6 vs. 62.8) found in the stress model (ES3). A p < 0.05 was considered statistically significant. chronic stress level in POST-intervention when compared to PRE- Statistical analysis was performed using SAS® v9.4 (SAS Institute, intervention (paired t-test, p = 0.01), while the variance increased. Cary, NC, United States). Similar to self-reported questionnaires, there is a significant effect related to the initial condition in the ESBSm results (p = 0.016), and no effect for age, BMI, or physical activity. 3 Results 3.1 Sample characteristics 3.3 Secondary outcomes The study sample consisted of 21 female participants who 3.3.1 Physiological variables completed the protocol in an average of 36 ± 11.4 days. The The bivariate analysis revealed some significant differences daily practice accomplishment was 62.28% (±19.1) regarding (Table 5), such as a decrease in mean respiratory rate from overall target home practice of 28 days. The mean age was 37.7 ± 0.259 to 0.222 Hz (i.e., 15.5 to 13.3 breaths per minute) and an 11.7 years (range: 19–59 years), with an average body mass index of increase in HRV-SDNN, HRV-LF/HF, and HRV-LFn, with no 24.2 ± 3.8 kg/m2. Most of the participants were nurses (61.9%), differences for HRV-RMSSD. followed by physicians (28.6%) and auxiliary nurses (9.5%). All No significant changes in PAT and peripheral temperature sociodemographic and clinical information is detailed in Table 3. parameters were observed. For EDA, the phasic component showed an increase (p = 0.012), whereas the average of the tonic component decreased POST intervention, although not significantly 3.2 Primary outcomes (p = 0.199). 3.2.1 Self-reported psychometric questionnaires 3.3.2 Acute stress measurement Figure 3 shows the classification PRE- and POST-intervention of Differences in physiological response measurement, each psychometric questionnaire according to the cut-off points of i.e., between the baseline measurement (BM), stress exposure previous studies (Aguiló Mir et al., 2021; Pedrozo-Pupo et al., 2020; (SE), and recovery measurement (RM), were analyzed to visualize Blevins et al., 2015; Kroenke et al., 2001). The classification is possible changes in the behavior of the signals during a stress- generally in three categories considering the degree of severity as induced task. All mean values and respective p-values could be seen low, moderate, or high, and in two categories for the PCL5 that in Supplementary Table S1. The most relevant variables are classifies those with high or low risk of developing PTSD. illustrated on the following plots. Some of them were log- In the PRE-intervention condition, the study sample presented transformed for a better visual representation and were indicated predominantly high anxiety trait scores (47.6%), moderate anxiety with (log) in the axis title. state (71.4%), and moderate perceived stress (71.4%) and showed The resting heart rate (HRV-HR) and pulse arrival time (PAT) increased risk of developing PTSD (71.4%). Regarding the showed significant differences in BM and RM stages when compared to depression level, most were classified within the low range SE (p < 0.001) under both conditions (Figures 5A, D, respectively). The (57.1%), while two individuals (9.5%) showed scores for severe HRV-SDNN changed its pattern, where BM differed from RM and SE depression. After intervention, all these profiles improved; the stages in the PRE-intervention condition, which was not seen after risk of PTSD reduced to 28.6%, the majority presented low intervention (Figure 5B). In the POST-intervention condition, SDNN symptoms of depression (81%), while the moderate group values at baseline increased significantly (p = 0.027). showed reduction from 33.3% to 19%. In addition, substantial Figure 5C shows the behavior of the power in low-frequency shifts from high to moderate anxiety and perceived stress can be band (HRV-LF), revealing differences between RM and SE at PRE- observed. intervention, which did not occur after intervention. Frontiers in Physiology 08 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 FIGURE 3 Severity classification of subjects based on the psychometrics cut-off points at PRE- and POST-intervention. (A) Stress perception with the Visual Analog Stress Scale (VASS), (B) risk to develop PTSD with Posttraumatic Stress Disorder Checklist for DSM-5 (PCL5), (C) perceived stress with the Perceived Stress Scale (PSS), (D) anxiety state with State Anxiety Inventory (STAI-S), (E) anxiety trait with Trait Anxiety Inventory (STAI-T), and (F) depression with Patient Health Questionnaire-9 (PHQ9). TABLE 4 PRE–POST psychometric questionnaire scores and corresponding statistical analysis. Psychometric questionnaire PRE POST Mean difference p-value VASS 66.57 (15.5) 31.95 (15.8) 34.62 <0.001 (a) PCL5 35.71 (16.9) 24.43 (16.4) 11.29 0.001 (b) PSS 19.05 (6.33) 15.43 (6.00) 3.62 0.023 (a) STAI-S 30.33 (9.48) 18.81 (7.43) 11.52 <0.001 (a) STAI-T 23.1 (10.6) 19.81 (8.90) 3.29 0.147 (b) PHQ9 9.85 (5.82) 6.47 (4.80) 3.38 0.003 (b) Key: Values represented as mean (SD). (a) Paired t-test. (b) Wilcoxon signed-rank. p< 0.05. Figure 6 shows the pattern of EDA and ST parameters. The tonic furthermore, values at SE and RM reduced significantly POST component (Tonic) of EDA showed differences between the BM and intervention (p = 0.047 and p = 0.021, respectively). The phasic the other stages (p < 0.001) in both conditions (Figure 6A); component (Phasic) and EDASymp differed between stages (p ≤ Frontiers in Physiology 09 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 0.001), but only Phasic showed difference in BM after intervention (p = 0.012). Regarding skin temperature, the gradient (TGradient) showed differences between the BM and RM stages in relation to the SE, in exception to RM vs. SE under POST condition (Figure 6D). No differences related to respiration were observed between stages (Figure 7A), although the respiratory rate was overall reduced after the intervention, significantly only in BM (p = 0.007). The ES3, the model to assess acute stress, did not reveal differences between stages, although the scores were all lower after intervention (Figure 7B). 4 Discussion Our study applied a new objective tool, the chronic stress FIGURE 4 Chronic stress level at PRE- and POST-intervention measured by multiparametric model (ESBSm), to assess the effectiveness of a 5- the multiparametric chronic stress model (ESBSm). Key: p < 0.05. session HRV-BF protocol for HCWs presenting mental health symptoms due to their work in the frontlines of the COVID-19 TABLE 5 Electrophysiological measure comparisons and corresponding statistical analysis. Parameters PRE POST p-value ECG HRV-HR, bpm 75.96 (9.48) 74.96 (10.01) 0.681 (a) HRV-SDNN, s 0.044 (0.013) 0.051 (0.016) 0.027 (a) HRV-RMSSD, s 0.012 (0.005) 0.012 (0.006) 0.801 (b) HRV-sdsd, s 0.012 (0.005) 0.012 (0.006) 0.801 (b) -2 HRV-VLF, s 0.118 (0.087) 0.173 (0.123) 0.047 (b) -2 HRV-LF, s 0.226 (0.190) 0.288 (0.182) 0.078 (b) -2 HRV-HF, s 0.139 (0.120) 0.129 (0.120) 0.175 (b) HRV-LF/HF ratio 204.77 (142.06) 413.15 (296.9) 0.003 (b) HRV-LFn, nu 53.62 (13.86) 66.24 (12.65) 0.003 (a) HRV-HFn, nu 46.38 (13.86) 33.76 (12.65) 0.003 (a) PAT PAT, ms 240.96 (10.29) 242.57 (12.97) 0.657 (a) stdPAT, ms 2.31 (1.52) 2.94 (3.29) 0.775 (b) Resp RR, Hz 0.259 (0.07) 0.222 (0.09) 0.007 (a) Pk, % 0.827 (0.04) 0.836 (0.06) 0.418 (a) EDA Tonic, µS −1.094 (0.285) −0.948 (0.487) 0.199 (b) Phasic, µS 0.038 (0.044) 0.099 (0.143) 0.012 (b) aucPhasic, µS 2.259 (2.644) 5.940 (8.594) 0.012 (b) EDASymp, µS 0.092 (0.170) 0.285 (0.627) 0.212 (b) ST TFinger, °C 30.57 (4.06) 31.44 (2.49) 0.334 (b) TGradient, °C 0.056 (0.058) 0.032 (0.073) 0.078 (b) 2 TPower, °C 951.20 (220.41) 995.25 (151.05) 0.334 (b) Key: Values represented as mean (SD). (a) Paired t-test. (b) Wilcoxon signed-rank. p < 0.05. Frontiers in Physiology 10 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 FIGURE 5 Boxplot of HRV parameters: (A) HR, (B) SDNN, (C) LF, and (D) PAT over the stages and conditions. Key: Light gray boxes show the significant differences between stages for each condition. (#) Differences between BM and the other stages (SE or RM). (+) Differences between RM and SE stages. p < 0.05. pandemic. The improvement in mental health is corroborated by lower helpful to stress management and to recover the autonomic scores on self-reported psychometric questionnaires after intervention, balance. Additionally, this improvement in respiratory rate reducing stress, risk of post-traumatic stress disorder, and symptoms of leads to better gas exchange and less chance of anxiety and depression. These results are consistent with those of hyperventilation, which often occurs in disorders such as previous studies that use similar instruments to assess the impact on anxiety (Leyro et al., 2021). Our analysis related to respiration mental health as published in this review (Di Nota et al., 2021), with during the stress-inducing tasks were compromised once the some studies in conjunction with an enhancement of certain tasks involved speech, so the data did not present the expected physiological biomarkers, such as heart rate variability. Thus, the increase in respiratory rate which was observed in other types of results obtained in the procedure using the ESBSm support the stressors, like cold or heat exposure (Tipton et al., 2017). hypothesis that a combined electrophysiological model provides An improvement in some HRV parameters, such as standard useful and robust information about mental health state and could deviation of normal beats (SDNN) and power in low frequency be considered to assess non-invasively changes over time. (LFn), was also observed. These parameters are often used to assess Furthermore, the acute stress model (ES3), applied to compare the effectiveness of biofeedback intervention as they are considered ANS physiological reactivity at baseline, stressor, and recovery biomarkers. SDNN varies especially due to RSA, which is mediated by measurements, showed an overall decrease after intervention, the parasympathetic system and precisely trained in HRV-BF, so an although nonsignificant. Unlike the chronic stress model, which increase in this parameter indicates a successful intervention (Limmer aims to find differences in cumulative stress, this model investigates et al., 2021). Furthermore, the LF power is influenced by both differences in acute physiological responses caused by stress-inducing sympathetic and parasympathetic branches of the ANS, but reflects tasks. As expected, there was an increase in the ES3 score when facing a baroreflex activity in resting conditions (McCraty and Shaffer, 2015). stressor and a reduction in resting periods, in both conditions. This Differences on the HRV-LFn may be related to the changes in behavior indicates that the intervention did not affect the acute stress respiratory rate since the low-frequency band (0.04–0.15 Hz) is response in the studied sample. influenced by breathing at 9 bpm or lower (Shaffer and Ginsberg, Regarding the electrophysiological variables, a reduction in 2017). In this sense, the increase in LF power is expected after HRV-BF the basal respiratory rate is observed, which indicates that the and corroborates the results shown in previous studies (Gross et al., participants were able to self-regulate their breathing, and this is 2016; Hasuo et al., 2020). Frontiers in Physiology 11 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 FIGURE 6 Boxplot of EDA and ST parameters: (A) Tonic, (B) Phasic, (C) EDASymp, and (D) TGradient, over the stages and conditions. Key: Light gray boxes show the significant differences between stages for each condition. (#) Differences between BM and the other stages (SE or RM). (+) Differences between RM and SE stages. p < 0.05. FIGURE 7 Boxplot of (A) respiratory rate (RR) and (B) ES3 acute stress model over the stages and conditions. Key: p < 0.05. The pulse arrival time, which is an index that varies inversely elicitation (Armañac et al., 2019; García Pagès et al., 2023). with blood pressure, showed a decrease during the stress-inducing Concerning the skin temperature, no differences were observed task when compared to baseline and recovery measurements. This after the intervention. All the parameters extracted from this pattern is in accordance with previous studies that demonstrated signal showed the same expected pattern, reducing during the this reactivity in healthy students during a similar protocol for stress stress exposure and recovering afterward (Pisanski et al., 2018). Frontiers in Physiology 12 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 The variations of electrodermal activity (EDA) in our sample our findings related to the effectiveness of HRV-BF for men. Second, may correspond only to central nervous system modulation related the relatively small sample size (n = 21) is analyzed, which could be to emotional and cognitive states of the individuals, once the reflected in statistically nonsignificant results and thus increasing the environmental temperature interference can be discarded as the risk of type II error (i.e., false negatives). skin temperature did not vary over time (Sarchiapone et al., 2018). The PRE–POST experimental design does not prevent us The phasic component of the EDA revealed differences after the from bias and limit the interpretation of the results. For this intervention. This component, which corresponds to faster- reason, we could not generalize the effectiveness of the applied changing activity, increased at baseline measurement, which intervention, which would be feasible in a case-control study, indicates an augmented sympathetic arousal, on the contrary to although, as an advantage, the present study design with paired what was expected after the intervention (Greco et al., 2021). We comparisons avoids the influence of interpersonal variations of hypothesize that the participants anticipated their stress responses the electrophysiological variables. In order to use the once they knew the elicitation tasks they were going to perform multiparametric chronic stress model (ESBSm) proposed by (i.e., Stroop test, arithmetic task, and description of feelings). Even Aguiló Mir et al. (2021), we assumed that the variability and so, it is important to highlight that both parameters extracted from severity of samples in both studies were similar and applied our the phasic component and the EDASymp were capable of standardized data to calculate the chronic stress level. The distinguishing between the stressor of basal and recovery necessary corrections were also made in EDA parameters due measurements. On the other hand, the tonic component of EDA to calibration problems related to the measuring device. showed an awaited pattern, increasing during the stress task, and a There was no follow-up to observe if the effects remained long- slower recovery time, seeming to accelerate slightly after the term after withdrawal of biofeedback. Further studies should be intervention, but still differing significantly from the baseline considered to assess the durability of the effects. stage. During the stress exposure and recovery stages, the tonic level reduced significantly after intervention, which suggests an improvement in psychophysiological state and autonomic 5 Conclusion regulation (Marzi et al., 2021). To summarize, the multiparametric chronic stress model Our study demonstrates that a 5-session HRV-BF protocol comprises information of a combination of the most relevant is an effective intervention to reduce stress, the risk of post- physiological variables altogether, and their scores were in traumatic stress disorder, and symptoms of anxiety and accordance with psychometric gold standards, therefore depression among our group of frontline HCWs who worked demonstrating to be a simple and objective tool to assess the during the COVID-19 pandemic. Our findings support that the effectiveness of a specific therapy. This procedure can be chronic multiparametric model based on electrophysiological especially useful and easily applied in clinical practices, thus variables provides relevant information about the current allowing to plan and properly combine different types of mental health status, being a reliable objective tool to assess interventions for a more assertive treatment. the effectiveness of HRV-BF and may be useful for other stress- In the current study, the HCWs’ adherence was reduction interventions and contexts of burden in mental satisfactory, and they reported feeling motivated throughout health. The variance increase in the model’s scores after the the biofeedback protocol, which reinforces the feasibility of intervention indicates that the effectiveness of the intervention the intervention in this context and the importance of depends on each individual; therefore, a systematic implementation by the health facilities as a preventive measurement should be crucial for follow-up and treatment program. Moreover, HRV-BF is a proactive intervention decisions. Further research should examine the robustness of that easily allows a multimodal measurement of these findings controlling possible confounders and, in electrophysiological parameters, ensuring the feasibility of addition, could investigate the reliability of wearables in the chronic stress model to assess its effectiveness and, providing feedback and monitoring physiological responses, more extensively, of other stress-reducing interventions. which would constitute a more unobtrusive and easier-to-use Further research could replicate the proposed method in instrument to reproduce the proposed procedure. case-control studies and other interventions to confirm its efficacy for particular applications. To make the procedure even easier and unobtrusive, a remote wearable device may be Data availability statement used to monitor the mental status in real-time and to assess the effectiveness of interventions more extensively. A remote The raw data supporting the conclusion of this article will be monitoring device would allow a longer follow-up and could made available by the authors, without undue reservation. be helpful in investigating other sources of stress and identifying protective and risk factors. Ethics statement 4.1 Limitations The studies involving human participants were reviewed and approved by the Ethical Committee of the Hospital Clínic of One limitation of this study is that the recruited sample Barcelona. The patients/participants provided their written comprises only women; therefore, it is not possible to extrapolate informed consent to participate in this study. Frontiers in Physiology 13 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 Author contributions Conflict of interest The concept and design of the study were conducted by TCR, Since 16th June 2022, the co-author PSS has been employed by SA, EGP and JA. TCR, PSS and SA recruited the volunteers, carried Frontiers Media SA. PSS declared his/her affiliation with Frontiers, out the experiments, and collected the data. Critical revision and and the handling Editor states that the process nevertheless met the contribution of scientific knowledge contribution were provided by standards of a fair and objective review. BL-B, SA and JA. The relevant data analysis and statistical analysis The remaining authors declare that the research was conducted were performed by TCR, PSS and LB. TCR wrote the manuscript in the absence of any commercial or financial relationships that with contributions from SA, EGP, PSS and JA. All authors could be construed as a potential conflict of interest. contributed to the article and approved the submitted version. Publisher’s note Funding All claims expressed in this article are solely those of the authors and This work was supported by the Fondo Supera COVID, funded by do not necessarily represent those of their affiliated organizations, or the Santander Bank, Conference of Rectors of Spanish Universities those of the publisher, the editors, and the reviewers. Any product that (CRUE) and Spanish National Research Council (CSIC). may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Acknowledgments Supplementary material We would like to thank healthcare workers all over the world for tirelessly fighting the coronavirus on the frontline, and especially The Supplementary Material for this article can be found online those from the Hospital Clínic of Barcelona and Hospital del Mar for at: https://www.frontiersin.org/articles/10.3389/fphys.2023.1147260/ collaborating with the scientific study carried out. full#supplementary-material References Aguiló Mir, S., García Pagès, E., López Barbeito, B., Castro Ribeiro, T., Garzón-Rey, J. M., Flores, E. C., Fuhr, D. C., Bayer, A. M., Lescano, A. G., Thorogood, N., and Victoria, S. and Aguiló, J. (2021). Design and validation of an electrophysiological based tool to assess (2018). Mental health impact of social capital interventions: A systematic review. Soc. chronic stress. Case study: Burnout syndrome in caregivers. Stress 24 (4), 384–393. doi:10. Psychiatry Psychiatric Epidemiol. 53 (2), 107–119. doi:10.1007/s00127-017-1469-7 1080/10253890.2020.1807512 García Pagès, E., Arza, A., Lazaro, J., Puig, C., Castro, T., Ottaviano, M., et al. (2023). Ahmed, A., Ali, N., Anna, G., Abd-alrazaq, A. A., Aziz, S., and Househ, M. (2021). Psychosomatic response to acute emotional stress in healthy students. Front. Physiology Mobile applications for mental health self-care: A scoping review. Comput. Methods 13, 2605. doi:10.3389/FPHYS.2022.960118 Programs Biomed. Update 1, 100041. doi:10.1016/j.cmpbup.2021.100041 Garzón-Rey, J. M., Lázaro, J., Milagro, J., Gil, E., Aguiló, J., and Bailón, R. (2017). Ahmed, T., Qassem, M., and Kyriacou, P. A. (2022). Physiological monitoring of Respiration-guided analysis of pulse and heart rate variabilities for acute emotional stress and major depression: A review of the current monitoring techniques and stress assessment. Comput. Cardiol. 44, 5–8. doi:10.22489/CinC.2017.318-093 considerations for the future. Biomed. Signal Process. Control 75, 103591. doi:10. Garzón-Rey, J. M. (2017). Psychosomatic approach to stress measurement” TDX (tesis 1016/j.bspc.2022.103591 doctorals en xarxa), Barcelona: Universitat Autònoma de Barcelona. Aristizabal, J. P., Navegantes, R., Melo, E., and Pereira, A. (2020). Use of heart rate Gil, E., Mendez, M., Jose Maria VergaraCerutti, S., Bianchi, A. M., and Laguna, P. variability biofeedback to reduce the psychological burden of frontline healthcare (2009). Discrimination of sleep-apnea-related decreases in the amplitude fluctuations of professionals against COVID-19. Front. Psychol. 11, 572191. doi:10.3389/fpsyg.2020. PPG signal in children by HRV analysis. IEEE Trans. Biomed. Eng. 56 (4), 1005–1014. 572191 doi:10.1109/TBME.2008.2009340 Armañac, P., Kontaxis, S., Lazaro, J., Laguna, P., Bailón, R., and Gil, E. (2019). Giorgi, G., Isaia Lecca, L., Alessio, F., Bondanini, G., Ginevra Lulli, L., Arcangeli, G., et al. Cardiovascular changes induced by acute emotional stress estimated from the pulse (2020). COVID-19-Related mental health effects in the workplace: A narrative review. Int. transit time difference” in 2019 computing in cardiology conference (CinC), Comput. J. Environ. Res. Public Health 17 (21), 7857–7922. doi:10.3390/ijerph17217857 Cardiol. 46, 1. doi:10.22489/CinC.2019.383 Greco, A., Valenza, G., Lanata, A., and Luca, C. (2016). CvxEDA: A convex Arza, A., Mario Garzón-Rey, J, Lázaro, J, Gil, E, Lopez-Anton, R, de la Camara, C, optimization approach to electrodermal activity processing. IEEE Trans. Biomed. et al. 2018. “Measuring acute stress response through physiological signals: Towards a Eng. 63 (4), 797–804. doi:10.1109/TBME.2015.2474131 quantitative assessment of stress.” Med. Biol. Eng. Comput., 57, 271. doi:10.1007/ s11517-018-1879-z Greco, A., Valenza, G., Lazaro, J., Mario Garzon-Rey, J., Aguilo, J., De-la-Camara, C., et al. (2021). Acute stress state classification based on electrodermal activity modeling. Billman, G. E. (2013). The LF/HF ratio does not accurately measure cardiac IEEE Trans. Affect. Comput. 1, 788–799. doi:10.1109/TAFFC.2021.3055294 sympatho-vagal balance. Front. Physiology. Billman 4, 26. doi:10.3389/fphys.2013.00026 Gross, M. J., Shearer, D. A., Bringer, J. D., Hall, R., Cook, C. J., and Kilduff, L. P. Blevins, C.A., Weathers, F. W., Davis, M. T., Witte, T. K., and Domino, J. L. (2015). (2016). Abbreviated resonant frequency training to augment heart rate variability and The posttraumatic stress disorder checklist for DSM-5 (PCL-5): Development and enhance on-demand emotional regulation in elite sport support staff. Appl. initial psychometric evaluation. J. Trauma. Stress 28 (6), 489–498. doi:10.1002/jts.22059 Psychophysiol. Biofeedback 41 (3), 263–274. doi:10.1007/s10484-015-9330-9 Cairns, P., Aitken, G., Pope, L. M., Cecil, J. E., Cunningham, K. B., Ferguson, J., et al. (2021). Hasuo, H., Kanbara, K., and Fukunaga, M. (2020). Effect of heart rate variability Interventions for the well-being of healthcare workers during a pandemic or other crisis: biofeedback sessions with resonant frequency breathing on sleep: A pilot study among Scoping review. BMJ Open 11 (8), e047498. doi:10.1136/bmjopen-2020-047498 family caregivers of patients with cancer. Front. Med. 7, 61. doi:10.3389/fmed.2020. Diez-Quevedo, C., Rangil, T., Sanchez-Planell, L., Kurt, K., and RobertSpitzer, L. 00061 (2001). Validation and utility of the patient health questionnaire in diagnosing mental Hayano, J., and Yuda, E. (2019). Pitfalls of assessment of autonomic function by heart disorders in 1003 general hospital Spanish inpatients. Psychosom. Med. 63 (4), 679–686. rate variability. J. Physiological Anthropol. 38 (1), 3–8. doi:10.1186/s40101-019-0193-2 doi:10.1097/00006842-200107000-00021 Heiss, S., Vaschillo, B., Vaschillo, E. G., Alix Timko, C., and Hormes, J. M. (2021). Di Nota, P. M., Bahji, A., and Groll, D. (2021). Proactive psychological programs Heart rate variability as a biobehavioral marker of diverse psychopathologies: A review designed to mitigate posttraumatic stress injuries among at-risk workers: A systematic and argument for an ‘ideal range. Neurosci. Biobehav. Rev. 121, 144–155. doi:10.1016/j. review and meta-analysis. Syst. Rev. 10 (1), 126. doi:10.1186/s13643-021-01677-7 neubiorev.2020.12.004 Frontiers in Physiology 14 frontiersin.org Castro Ribeiro et al. 10.3389/fphys.2023.1147260 Hernando, A., Lázaro, J., Gil, E., Arza, A., Garzón, J. M., López-Antón, R., et al. (2016). Pappa, S., Ntella, V., Giannakas, T., Giannakoulis, V. G., Papoutsi, E., and Inclusion of respiratory frequency information in heart rate variability analysis for stress Katsaounou, P. (2020). Prevalence of depression, anxiety, and insomnia among assessment. IEEE J. Biomed. Health Inf. 20 (4), 1016–1025. doi:10.1109/JBHI.2016. healthcare workers during the COVID-19 pandemic: A systematic review and meta- 2553578 analysis. Brain Behav Immun 88, 901–907. doi:10.1016/j.bbi.2020.05.026 Kennedy, L., and Parker, S. H. (2019). Biofeedback as a stress management tool: A Pedrozo-Pupo, T., Carlos, J., and Campo-Arias, A. (2020). Depression, perceived systematic review. Cognition, Technol. Work 21 (2), 161–190. doi:10.1007/s10111-018- stress related to COVID, post-traumatic stress, and insomnia among asthma and COPD 0487-x patients during the COVID-19 pandemic. Chronic Respir. Dis. 17, 147997312096280. doi:10.1177/1479973120962800 Kim, H. G., Cheon, E. J., DaiBai, S., YoungLee, H., and Koo, B. H. (2018). Stress and heart rate variability: A meta-analysis and review of the literature. Psychiatry Investig. Pisanski, K., Kobylarek, A., Jakubowska, L., Nowak, J., Walter, A., Błaszczyński, K., 15, 235–245. doi:10.30773/pi.2017.08.17 et al. (2018). Multimodal stress detection: Testing for covariation in vocal, hormonal and physiological responses to trier social stress test. Hormones Behav. 106, 52–61. Kirschbaum, C., Pirke, K-M., and DirkHellhammer, H. (1993). The ‘trier social stress doi:10.1016/j.yhbeh.2018.08.014 test’ – a tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology 28 (1–2), 76–81. doi:10.1159/000119004 Pizzoli, S. F. M., Marzorati, C., Gatti, D., Monzani, D., Mazzocco, K., and Pravettoni, G. (2021). A meta-analysis on heart rate variability biofeedback and depressive Kroenke, K., Spitzer, R. L., and Williams, J. B. W. (2001). The PHQ-9 validity of a brief symptoms. Sci. Rep. 11 (1), 6650–6710. doi:10.1038/s41598-021-86149-7 depression severity measure. J. General Intern. Med. 16 (9), 606–613. doi:10.1046/j. 1525-1497.2001.016009606.x Posada-Quintero, G., Hugo, F., Florian, J. P., Orjuela-Cañón, Á. D., and Chon, K. H. (2016). Highly sensitive index of sympathetic activity based on time-frequency spectral Lázaro, J., Alcaine, A., Romero, D., Gil, E., Laguna, P., Pueyo, E., et al. (2014). analysis of electrodermal activity. Am. J. Physiology - Regul. Integr. Comp. Physiology Electrocardiogram derived respiratory rate from QRS slopes and R-wave angle. Ann. 311 (3), R582–R591. doi:10.1152/ajpregu.00180.2016 Biomed. Eng. 42 (10), 2072–2083. doi:10.1007/s10439-014-1073-x Purtle, J., Nelson, K. L., Counts, N. Z., and Yudell, M. (2020). Population-based Lazaro, J., Gil, E., Jose Maria, V, and Laguna, P. (2014). Pulse rate variability analysis approaches to mental health: History, strategies, and evidence. Annu. Rev. Public Health for discrimination of sleep-apnea-related decreases in the amplitude fluctuations of 41 (1), 201–221. doi:10.1146/annurev-publhealth-040119-094247 pulse photoplethysmographic signal in children. IEEE J. Biomed. Health Inf. 18 (1), 240–246. doi:10.1109/JBHI.2013.2267096 Remor, E. (2006). Psychometric properties of a European Spanish version of the perceived stress scale (PSS). Span. J. Psychol. 9 (1), 86–93. doi:10.1017/ Lehrer, P. (2013). How does heart rate variability biofeedback work? Resonance, the S1138741600006004 baroreflex, and other mechanisms. Biofeedback 41 (1), 26–31. doi:10.5298/1081-5937- 41.1.02 Sarchiapone, M., Gramaglia, C., Iosue, M., Carli, V.r, Mandelli, L., Serretti, A., et al. (2018). The association between electrodermal activity (EDA), depression and suicidal Lehrer, P., Kaur, K., Sharma, A., Shah, K., Huseby, R., Bhavsar, Jay, et al. (2020). Heart behaviour: A systematic review and narrative synthesis. BMC Psychiatry 18 (1), 22–27. rate variability biofeedback improves emotional and physical health and performance: doi:10.1186/s12888-017-1551-4 A systematic review and meta analysis. Appl. Psychophysiol. Biofeedback 45 (3), 109–129. doi:10.1007/s10484-020-09466-z Scavone, A., Kadziolka, M. J., and CarlinMiller, J. (2020). State and trait mindfulness as predictors of skin conductance response to stress. Appl. Psychophysiol. Biofeedback 1, Lehrer, P., Vaschillo, B., Zucker, T., Graves, J., Katsamanis, M., Aviles, M., et al. 221–228. doi:10.1007/s10484-020-09467-y (2013). Protocol for heart rate variability biofeedback training. Biofeedback 41 (3), 98–109. doi:10.5298/1081-5937-41.3.08 Schuman, D. L., and Killian, M. O. (2019). Pilot study of a single session heart rate variability biofeedback intervention on veterans’ posttraumatic stress symptoms. Appl. Lesage, F.- X., Berjot, S., and Deschamps, F. (2012). Clinical stress assessment using a Psychophysiol. Biofeedback 44 (1), 9–20. doi:10.1007/s10484-018-9415-3 visual Analogue scale. Occup. Med. 62 (8), 600–605. doi:10.1093/occmed/kqs140 Shaffer, F., and Ginsberg, J. P. (2017). An overview of heart rate variability metrics and Leyro, Teresa M., VersellaMin-Jeong, Y., Mark, V, Hannah, R. B., Hoyt, D. L., and norms. Front. Public Health 5, 258. doi:10.3389/fpubh.2017.00258 Paul, L. (2021). Respiratory therapy for the treatment of anxiety: Meta-analytic review and regression. Clin. Psychol. Rev. 84, 101980. doi:10.1016/j.cpr.2021.101980 Shaffer, F., and Meehan, Z. M. (2020). A practical guide to resonance frequency assessment for heart rate variability biofeedback. Front. Neurosci. 14, 570400. doi:10. Limmer, A., Laser, M., and Schütz, A. (2021). Mobile heart rate variability biofeedback 3389/fnins.2020.570400 as a complementary intervention after myocardial infarction: A randomized controlled study. Int. J. Behav. Med. 3, 230–239. doi:10.1007/s12529-021-10000-6 Sobregrau Sangrà, P., Aguiló Mir, S., Aguiló MirCastro Ribeiro, T., Esteban- Sepúlveda, S., García Pagès, E., et al. (2022). Mental health assessment of spanish LinMei, I., Fan, S. Y., Yen, C. F., Yeh, Y. C., Tang, T C., Huang, M. F., et al. (2019). Heart rate healthcare workers during the SARS-CoV-2 pandemic. A cross-sectional study. variability biofeedback increased autonomic activation and improved symptoms of Comprehensive psychiatry 112 (January) : 152278. doi:10.1016/j.comppsych.2021. depression and insomnia among patients with major depression disorder. Clin. 152278 Psychopharmacol. Neurosci. 17 (2), 222–232. doi:10.9758/cpn.2019.17.2.222 Spielberger, C. D., Gorsuch, R. L., Lushene, R. E., Vagg, P. R., and Jacobs, G. A. (1983). Martínez, J. P., Almeida, R., Olmos, S., Rocha, A. P., and Laguna, P. (2004). A wavelet- Manual for the state-trait anxiety inventory. Palo Alto, CA: Consulting Psychologists based ECG delineator: Evaluation on standard databases. IEEE Trans. Biomed. Eng. 51 Press. (4), 570–581. doi:10.1109/TBME.2003.821031 Stroop, J. R. (1935). Studies of interference in serial verbal reactions. J. Exp. Psychol. 18 Marzi, C., Greco, A., and Vanello, N. (2021). Towards a model of arousal change after (6), 643–662. doi:10.1037/h0054651 affective word pronunciation based on electrodermal activity and speech analysis. Biomed. Signal Process. Control 67, 102517. doi:10.1016/j.bspc.2021.102517 Tipton, M. J., Abbi Harper, J. F. R. P., Costello, J. T., and Costello, J. T. (2017). The human ventilatory response to stress: Rate or depth? J. Physiology 595 (17), 5729–5752. Mateo, J., and Laguna, P. (2003). Analysis of heart rate variability in the presence of doi:10.1113/JP274596 ectopic beats using the heart timing signal. IEEE Trans. Biomed. Eng. 50 (3), 334–343. doi:10.1109/TBME.2003.808831 Tonhajzerova, I., Ondrejka, I., Turianikova, Z., Javorka, K., Calkovska, A., and Javork, M. (2012). Heart rate variability: An index of the brain-heart interaction. Tachycardia. McCraty, R., and Shaffer, F. (2015). Heart rate variability: New perspectives on InTech. doi:10.5772/25763 physiological mechanisms, assessment of self-regulatory capacity, and health risk. Glob. Adv. Health Med. 4 (1), 46–61. doi:10.7453/gahmj.2014.073 World Health Organization (2022). World mental health report: Transforming mental health for all. Geneva, Switzerland: World Health Organization. Moss, D., and Shaffer, F. (2017). The application of heart rate variability biofeedback to medical and mental health disorders. Biofeedback 45 (1), 2–8. doi:10.5298/1081- Yasuma, F., and Ichiro Hayano, J. (2004). Respiratory sinus arrhythmia: Why does the 5937-45.1.03 heartbeat synchronize with respiratory rhythm?” Chest. Elsevier 125 (2), 683–690. doi:10.1378/chest.125.2.683 O’Rourke, N., and Hatcher, L. (2013). A step-by-step approach to using SAS for factor analysis and structural equation modeling. Cary, NC: SAS Institute Inc., 412. doi:10. Yu, B., Funk, M., Hu, J., Wang, Q., and Feijs, L. (2018). Biofeedback for everyday stress 5555/2530480 management: A systematic review. Front. ICT 5, 23. doi:10.3389/fict.2018.00023 Frontiers in Physiology 15 frontiersin.org