A total of 75 participants were recruited to the study, including 44 (59%) women and 31 (41%) men, with an average age of 23.5 years (SD 6.39; range, 18-55 years). The sample predominantly identified themselves as white British (n=55, 73%) and the remainder, as Asian (n=8, 11%), white other (n=4, 5%), African (n=2, 3%), black British (n=1, 1%), Greek (n=1, 1%), Hispanic (n=1, 1%), mixed Hispanic (n=1, 1%), and Mauritian (n=1, 1%). All participants received a £15 gift voucher for their time. The study was approved by the ethics committee of the University of Surrey, UK, and written informed consent was obtained from all participants (identifier: 1190-PSY-16).

During the stressor phase of the study, participants listened to a 2-min audio recording that described a situation where they had been falsely accused of stealing a purse and had to defend themselves to the police; this stressor has been used in previous studies (eg, [30]). The participants were instructed to imagine themselves in the scenario and were given 2 min to mentally think and prepare a statement for their defense in front of a video recorder. The recording stated that their speech would be evaluated and marked by the experts for fluency and confidence. This aimed to instill a social evaluative threat that has been shown to increase stress and rumination and induce physiological reactivity [31,32]. However, in reality, the speeches were not evaluated, as the performance ratings were not relevant to this study.

Participants in the BioBase condition were guided through a clinically validated version of the Papworth-Benson breathing exercise [33]. Originally developed at Papworth Hospital in Cambridgeshire, UK, more than 5 decades ago, this breathing method focuses on diaphragmatic breathing to use the full lung capacity and slow down breathing as well as trigger vagus nerve stimulation and consequently, the relaxation response. Through the process of photoplethysmogram and the use of the camera and flash on the mobile phone, the app measured changes in the user’s heart rate during a guided breathing exercise and displayed their heart wave on the mobile phone screen in real time. Participants in the control and rumination conditions used a modified version of the Biobase app during the intervention period, which measured their heart rate, but did not guide them through the breathing exercise.

After entering the laboratory, participants were asked to read the information sheet explaining the study. After providing written consent, the participants completed a demographic questionnaire. They were subsequently fitted with electrodes and acclimatized to the Biopac amplifier. The Biopac amplifier recorded heart data continuously throughout the experiment. All participants then relaxed for 6 min while the experimenter left the room to minimize distraction. On resuming the exercise, the participants rated their level of stress using a 7-point visual analogue scale (1, very little pressure to 7, extreme pressure). Thereafter, the participants listened to the audio stressor recording for 2 min, as described above in the Speech Task section, and were given 2 min to prepare a statement for their defense. They then presented their statement for 3 min in front of the experimenter and a video camera. Subsequently, the participants rated their level of stress using the 7-point stress scale. Following the speech task, participants were randomly assigned using a random number-generator program [39] to one of the three groups (control, BioBase, or rumination) for 6 min. As this study was not a preregistered randomized controlled trial, it was considered a pilot study.

Participants assigned to the breathing condition used the BioBase app on the mobile phone and followed the guided breathing instructions. Individuals in the rumination condition were asked to think and reflect about their speech and focus on performance-based thoughts and were guided during this task by a modified version of the Thoughts Questionnaire [35]. Participants in the control condition were asked to sit passively. At the end of the intervention period, participants rated their level of stress. All participants then completed the original Thoughts Questionnaire [35] to measure the actual levels of rumination during the intervention phase. Finally, participants were disconnected from the equipment, fully debriefed, and given a gift voucher.

All data were screened for measurement artifacts using the ARTiiFACT software [40] by one of the authors (SS) who was blinded to the conditions. Artifacts were identified using the algorithm developed by Berntson and colleagues [41], and a distribution-based threshold value calculated individually for each participant. Flagged beat intervals were visually checked and, if confirmed as artifacts, deleted and substituted by means of cubic spline interpolation of neighboring intervals. HRV analysis is typically performed in time or frequency domains. The time-domain measure root mean square successive difference (RMSSD) is considered to indicate vagally mediated HRV; values < 50 ms are considered unhealthy, values > 100 are considered healthy, and values in between indicate risk. The frequency-domain analysis high frequency (HF), which is considered to depict parasympathetic regulation, was also calculated for each person. Measurements and analyses followed established guidelines [42]. For the analysis, we grouped the HRV data into baseline, stressor, and intervention groups and analyzed 6-min segments, which are sufficient for HRV analysis [43]. Following the standard procedure, data were screened, outliers above and below 3 SDs were excluded, and data were transformed where necessary.

To examine the effects of the intervention on cardiac recovery in the poststress period, a between-subjects analysis of covariance for group (Biobase, rumination, and control) was conducted using the recovery measures as the outcome and the reactivity change score as a covariate. As individual differences in stress response could influence the findings, following previous research [9], we treated reactivity as a covariate to statistically remove the influence of reactivity on recovery. Apart from heart rate, a significant effect of group, with medium-to-large effect size, was found for RMSSD, pNN50, and HF. In addition, in all cases, the recovery rate was higher in the BioBase group than in the rumination and control groups. There were no significant differences between the rumination and control groups. Thus, individuals who used the breathing app demonstrated enhanced recovery from stress. The results of the analysis are presented in Table 1.

As stress is an unavoidable part of life and the workplace becomes increasingly pressured, providing fewer opportunities for recovery, it is important to identify easy-to-use practical aids that can help individuals unwind and recover after the stressful period. The main finding of this study was that physiological recovery following stress exposure was enhanced in individuals who were guided through the BioBase breathing condition. This is an interesting finding because higher HRV is associated with reduced cardiovascular risks and other health outcomes [18,20,44].

We examined the effects of rumination on recovery in this study. Cognitive rumination has been defined as repetitive and intrusive negative thinking about past stressor(s) [45], and rumination is one of the main contributing factors that disrupts poststress recovery [46,47]. We were successful in manipulating rumination in the laboratory; however, rumination in the present study was not associated with reduced HRV, as there was no significant difference in poststress recovery between the rumination and control groups (as compared to [48]), although both groups showed reduced HRV as compared to the BioBase group.

Although there was a difference between the groups with respect to physiological parameters during recovery, there was no difference in subjective stress, as the levels of perceived stress returned to baseline for all groups in the recovery period. This finding is intriguing and suggests that perceptions of and actual physiology are not always in accordance. Individuals may think they have recovered from the effects of stress but may be unaware of their actual physiological state. This may not be an issue in the short term, but a mismatch between one’s psychological and physiological state could lead to long-term health issues over time, especially in individuals who are exposed to high levels of stress or work in demanding jobs. Interventions aimed at increasing sensitivity to one’s internal body state—perhaps, by enhancing interoceptive awareness—are needed. Somatic, or interoceptive awareness, could be defined as one’s perceptual awareness of signals from the inside the body, such as heartbeats, breathing, bowel movements, and higher-order processes including beliefs and attitudes as well as emotions regarding those perceptions [49-51]. Interoceptive awareness can be measured and trained. Such training might rely on the fact that simply noticing feelings, particularly bodily sensations associated with particular feelings, can foster emotional regulation. This is an aspect of mindfulness training that has been previously explored in studies of meditation training on interoception [52,53]. In the future, studies should focus on the integration of multimodal systems such as digital screen-based devices (mobile phones) and multidimensional theoretical models of interoception that take into account measurement of autonomic nervous system disruption using HRV data in order to build multimodal, digital therapeutic systems that measure interoception as well as offer training experiences that might help manage anxiety-based disorders embedded in apps such as BioBase.

There are some caveats with this present study. Heart rate monitoring was stopped after 6 min; therefore, we do not know how long the actual effect on HRV lasted. It will be of interest for future studies to establish the benefits of guided breathing over time and examine whether the effect fades, stabilizes, or increases with repetition. Repeated use of the guided breathing app may help individuals who chronically experience stress to recover more quickly physiologically and improve their health outcomes and functioning. Although physical fitness has been associated with enhanced physiological recovery from stress [7,54], we did not assess the levels of physical fitness in this study, which could be perceived as a limitation. However, there was no significant difference between the groups in terms of self-reported physical activity, and therefore, it seems extremely unlikely that physical fitness could account for the difference between the conditions in HRV recovery. Nonetheless, future research should control for levels of fitness to completely rule out the possibility that fitness played a key role in our findings.

Given the ubiquitous nature of stress and the documented evidence associating stress with cardiovascular disease and mortality, it is important to identify affordable and practical aids that help individuals unwind and recover after the stressful period. To our knowledge, this is the first study to examine the effects of the BioBase breathing app on HRV under controlled conditions and provide evidence for its efficacy in aiding poststress recovery. These findings also provide evidence of the utility of biofeedback breathing in augmenting physiological recovery and suggest that app-based breathing interventions are effective in reducing cardiovascular disease risk.