Physical inactivity is one of the major risk factors for global mortality; therefore, an active lifestyle is highly important to promote health and longevity. The risk for cardiovascular diseases, type 2 diabetes, osteoporosis, metabolic syndrome, specific cancers, and several other diseases is reduced by being physically active [1], and physical exercise can be prescribed as a medicine for 26 common chronic diseases [2]. In addition, mental health is positively affected by regular physical activity. Physical exercise induces the preservation of brain volume, which is associated with cognitive benefits [3,4]; it promotes blood-brain barrier integrity and protects against central nervous system infiltration of immune cells [5], and neurodegenerative diseases, such as Alzheimer and Parkinson disease, are counteracted by an active lifestyle [6].

To show the influence of exercise type, frequency, duration, and intensity on mental health, a cross-sectional study was conducted with 1.2 million individuals in the United States between 2011 and 2015 [7]. It was shown that regularly physically active persons have approximately half as many days of poor mental health than physically inactive persons. Furthermore, it was demonstrated that any kind of exercise had a reduced mental health burden compared with not exercising. A duration of at least 45 minutes with a frequency of 3 to 5 times a week is recommended. However, the authors also cautioned that more exercise is not always better [7].

To improve peak oxygen uptake (VO₂peak), it has been shown that individuals who perform interval training have a higher increase in VO₂peak [8] or have a shorter exercise duration with similar increases in VO₂peak compared with those who perform continuous exercise [9-11]. The comparison between low- and high-fit groups showed that the high-fit participants reacted and recovered faster within and after a psychological stress situation [12,13]. Furthermore, high-fit individuals had a lower cortisol concentration at rest and higher variability during the stressor compared with low-fit individuals. Moreover, a low cardiorespiratory fitness at the age of 18 years is associated with serious depression in later life [14], whereas implementing regular physical exercise training induces comparable efficacy in the treatment of major depression as antidepressant medication [15]. The described positive effects of physical exercise on mental health are accompanied by increased overall resilience [16].

Currently, there is no uniform and stringent definition of resilience [17-19]. A common psychological definition of resilience is “a person’s ability to adapt successfully to acute stress, trauma or more chronic forms of adversity” [20]. According to Kalisch et al [17], it is defined as the “empirically observable phenomenon [...] that someone does not develop lasting mental health problems although he or she is subject to adversity.” Recent resilience theories suggest a distinction to be made between resilience as an outcome, that is, the change in mental health relative to the stressor burden in a certain period, and resilience mechanisms, that is, variables that directly affect resilient outcomes [17]. By now, several psychosocial characteristics of a resilient phenotype (eg, lower levels of denial, avoidant coping behavior, and high positive emotionality) and multiple ways of promoting it (eg, improving problem-solving and planning skills) have been described [16]. Emotion regulation ability is considered a potential resilience mechanism and finds prominence in dynamic resilience theories [17,21]. Previous cross-sectional studies have indicated a positive relationship between physical activity and emotion regulation [22,23]. However, the driving neural circuits and underlying molecular pathways in resilience, especially in the context of physical exercise, are only beginning to be unraveled [16].

The major neuroendocrine and neural drivers of the responses to stress are the hypothalamic-pituitary-adrenal (HPA) axis, which mediates glucocorticoid (cortisol) release, and the sympathetic nervous system, which initiates the release of catecholamines (epinephrine and norepinephrine) in interaction with the immune system. Adequate activation of these systems in response to psychological or physical stress and subsequent restoration of homeostasis is critical to physical and mental health [16,24]. Malfunction of this responsive orchestration is associated with the development of several chronic diseases, including psychiatric diseases, autoimmune diseases, and cardiovascular diseases, contributing to mortality worldwide [25-27]. Acute bouts of physical exercise activate the HPA axis and the sympathetic nervous system, leading to increased stress hormone release followed by immune cell mobilization in a dose-responsive fashion [28-30]. When cessation of the physical stressor and the ensuing reconstitution of a homeostatic state occurs after a moderate period, adaptive mechanisms may result in improved stress reactivity. These adaptations include neurogenesis, improved synaptic plasticity and neuroprotection (eg, by increased levels of brain-derived neurotrophic factor [30,31]), and the promotion of an anti-inflammatory state (eg, by enhanced anti-inflammatory cytokines and reduced c‑reactive protein release [32]). Consequently, regular physical exercise may increase resilience by promoting a more adaptive activation of the HPA axis. Similarly, an exercise-induced shift in the tryptophan pathway, resulting in decreased kynurenine and increased kynurenic acid plasma levels, is considered to promote neuroprotection and resilience [33,34]. Furthermore, physical exercise activates the endocannabinoid system, which is considered to result in cognitive benefits and an antidepressant effect [35-37]. Thus, exercise prescription constitutes a useful means for the prevention and treatment of psychological and neurodegenerative diseases and moreover offers the possibility to enhance resilience in the general population.

Structured physical exercise interventions have been successfully implemented to improve disease states in brain diseases, such as major depression, Alzheimer disease, and Parkinson disease [15,38,39]. However, personalized training in a face-to-face fashion requires a high workload for coaches, and participants need to adhere to a strict time schedule. Individualized web-based physical exercise interventions allow trainers to remotely supervise multiple participants in a time- and cost-effective manner, for example, by providing weekly feedback. In turn, participants are highly flexible when integrating the training sessions into their daily routines and still receive individualized training programs from a coach. Our research group had previously shown in different types of patients that a web-based exercise program is useful, does not harm any participant, and improves physiological parameters, such as VO₂peak [40,41], and clinically relevant parameters [42-45].

Physical activity is positively associated with blunted stress reactivity and shorter stress recovery in response to various stressors [46,47]. To investigate acute stress reactivity and recovery in a psychophysiological laboratory, an experimental paradigm that significantly induces stress at the psychological and physiological levels is needed. Dickerson and Kemeny [48] identified several elements that reliably elicit significant HPA axis responses in participants: (1) physical stressors (eg, heat or cold), (2) mental challenges (eg, arithmetic tasks and working memory tasks), and (3) social evaluation (eg, giving a speech in front of a jury). In psychological stress research, there are several established stress paradigms that make use of these elements. The Socially Evaluated Cold Pressor Task (SECPT) [49] and the ScanSTRESS-C [50] for acute stress induction are experimental paradigms that have previously been proven effective in eliciting acute stress responses at the psychological (ie, decreases in well-being), endocrine (ie, high levels of catecholamines and cortisol), and physiological (ie, higher heart rate levels) levels [50-52].

In this study, we implement an 8-week randomized controlled trial of a web-based, individualized physical exercise training intervention to evaluate the link between aerobic capacity and resilience in healthy untrained adults. Therefore, baseline values of performance diagnostics to determine the innate aerobic capacity (VO₂peak) as well as emotion regulation abilities, stress reactivity, and stress recovery (saliva cortisol) as potential resilience mechanisms are estimated. We hypothesize that (1) participants with a high innate aerobic capacity will show a higher resilience to stress compared with participants with low aerobic capacity. This will be followed by 8 weeks of individualized web-based physical exercise training combining continuous and interval-type running exercises with the aim of progressively increasing aerobic capacity. After completion of the exercise intervention, performance diagnostics will be repeated to estimate the acquired aerobic capacity, as well as to assess the psychological parameter. In addition, secondary outcomes of molecular factors, including hair cortisol levels, endocannabinoids, catecholamines, cytokines, and cell-free DNA, and results of several questionnaires will be determined throughout the study. We hypothesize that (2) the exercise intervention will improve the aerobic capacity of the intervention group, which (3) will be accompanied by improvements in resilience factors. This will offer the possibility to exploratively study the underlying molecular mechanism (stress hormone release and immune reactions) in the promotion of resilience by regular physical activity.

This study is designed as a prospective randomized controlled trial with 2 phases of data collection before (baseline examination; time point 0 [T0] and time point 1 [T1]) and after (final examination; time point 2 [T2] and time point 3 [T3]) 8 weeks of web-supervised physical exercise training intervention (T1-T2; Figure 1). The multidisciplinary single-center trial is a collaboration among the Institute of Physiological Chemistry, Department of Psychiatry and Psychotherapy, Department of Clinical Psychology and Neuropsychology, and Department of Sports Medicine, Disease Prevention and Rehabilitation. The medical association Rhineland-Palatinate approved the study (July 29, 2019, ID: 2019-14305), and the study was registered with the German Clinical Trials Register (DRKS; DRKS00018078; October 2, 2019) and conformed to the standards of the Declaration of Helsinki of the World Medical Association. Administrative changes of the protocol, if needed, are minor corrections and clarifications that have no effect on the way the study is to be conducted. An amendment will be approved by the Ethics Committee Landesärztekammer Rhineland-Palatinate before implementation. This protocol follows the guidelines from the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) [53] and the CONSORT-eHEALTH checklist (Multimedia Appendix 1) [54]. A SPIRIT diagram of enrollment, interventions, and assessments is provided in Figure 1.

The primary (VO₂peak and saliva cortisol) and key secondary outcomes, trial methods, and designs are summarized in the World Health Organization trial registration data set (Textbox 1). This study includes 4 time points of investigation of physiological and psychological parameters, including the primary end points of a change in VO₂peak and saliva cortisol from the baseline: at T0 and T1, baseline values will be estimated, and at T2 and T3, data collection following the control or exercise intervention will be performed. T0 and T2, as well as T1 and T3, are structured in the same way, except for the separately listed psychological examinations at the measurement points. T1 and T3 will be performed after a maximum of 10 days after T0 and T2, respectively. Written consent after a detailed oral explanation will be obtained from all participants at the beginning and after completion of the study by the responsible study physician. At the beginning of the study (T0), the participants will not be fully informed about the purpose and aim of the psychological stress tests. Full clarification will take place at T3. If a participant wishes to discontinue the study prematurely, they will be fully informed about the purpose and aim of the psychological stress tests.

Participant recruitment commenced in September 2019. Participants were and will be recruited via flyer announcements for the study (Multimedia Appendix 2) at multiple locations at the Johannes Gutenberg-University of Mainz and the University Medical Center campuses, as well as diverse places for leisure activities, as impersonal recruitment. Inclusion into the study is a two-step procedure beginning with a telephone survey according to the eligibility criteria (Textbox 2). At the first visit, sport capability will be confirmed and the participants will be included in the study. Blinding of participants will not take place. After telephone screening and recruitment, participants will be randomized (Figure 2) into the intervention group (IG) or control group (CG). The IG will participate in the 8-week sports intervention based on a combination of continuous and interval-type running exercises. Participants in the IG will be told not to engage in further high-intensity sports activities. Participants in the CG will be told not to change their daily lifestyles during the intervention period.

The included participants will be randomly assigned with a 50%-50% allocation to the IG or CG. Randomization will be performed using the EXCEPT NUMBER command in Microsoft Excel. In addition, randomization will be stratified by gender with an expected ratio of 50%-50% and BMI with an expected ratio of 33% overweight (BMI ≥25) and 67% normal weight (BMI <25) to recruit CG and IG participants with homogeneous baseline characteristics.

To work with a homogenous study population and to reduce the influence of different characteristics of the study population on study outcomes, the inclusion and exclusion criteria for participating in the study were chosen as listed in Textbox 2. It is known that both neural and behavioral processing of stress is altered in clinical samples with conditions such as depression [55] or anxiety disorders [56]. To minimize differences between participants, we excluded individuals with current or past psychological treatment as assessed in our standardized telephone screening. To reduce the impact of hormonal contraceptives in women on stress reactivity (saliva cortisol) and emotional memory [57], only women taking no hormonal contraceptives are included in our study. Alcohol abuse was defined according to the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) [58] as a regular or daily drinking routine.

The first visit (T0) includes the initial screening of the participant’s height and weight. A pulmonary function test will be used to determine vital capacity, forced expiratory volume in 1 second (Spirometry, Bodybox 5500, Medisoft Group) and, if applicable, to identify the presence of undiagnosed restrictive or obstructive respiratory abnormalities [59]. Furthermore, the cardiological values of resting blood pressure and resting electrocardiogram (Schiller AT-60) will be measured in a supine position to ensure proper cardiovascular function. To determine the long-term cortisol value over 2 months, a 2 cm–long strand of hair will be cut off at the back of the head directly from the skin [60]. A bioelectric impedance analysis (In Body 3.0, Biospace) and blood counts will be conducted during the second visit (T1) in the morning before the incremental running test after a minimum of 8 hours of overnight fasting. After the completion of the control or exercise intervention, the described parameters will be reassessed during the third or fourth visit (T2 or T3).

Owing to the multidimensional approach to stress resilience outlined above, we required a lot of information about the participants’ lifestyle as well as psychological and medical anamnesis with the help of several questionnaires. To minimize the total experimental time, participants will complete part of the questionnaires on the web before their arrival at T0. Further questionnaires will be completed within the resting and recovery periods during the on-site appointments.

To detect differences in the ability to regulate emotions, 2 different tests will be used in parallel to the stress tests.

To determine the individual VO₂peak as the primary end point and the training areas, participants will complete a stepwise incremental running test on a treadmill (Saturn, HP-Cosmos) until subjective exhaustion is reached or general indications for stopping an exercise test according to the American College of Sports Medicine guidelines are fulfilled [59]. Maximal oxygen uptake is reached if a plateau of VO₂ is observed within the final 2 work rates of the stepwise incremental running test [59]. Here, we have untrained participants who may not be able to reach a plateau in VO₂ uptake. Therefore, we will use VO₂peak and take the highest value of VO₂. To avoid outliers, there should be no high variance around this VO₂ value within the 4 previously taken breaths. The exercise test includes a starting velocity of 4 km/hour, a duration of each step of 3 minutes, a pause time between 30 seconds and 60 seconds, and an increase in velocity by 1.5 km/hour from step to step. To determine the lactate threshold, 20 µL of whole blood will be collected from the fingertip with an end-to-end capillary (Sodium-Heparin, EKF-Diagnostics GmbH) at rest and between the stages. Heart rate (electrocardiogram), as well as oxygen uptake and carbon dioxide release (spiroergometry), will be continuously recorded. The latter will be used to estimate VO₂peak for IG and CG at T1 and T3. VO₂peak at T1 will be considered as innate aerobic capacity, and VO₂peak of the IG at T3 will be considered as acquired aerobic capacity. Absolute and relative contraindications will be defined according to the American College of Sports Medicine guidelines [59].

Infrared thermography is a noninvasive, contact-free, nonradiating tool used to measure surface radiation temperature and patterns [107]. In this study, a high-resolution infrared camera (Jenoptik VarioCam; 480×640 pixels; focal plane array; spectral range 7.5-14 micrometers; accuracy ± 0.05 K; and set emissivity: 0.98) will be applied for the measurement of mean surface radiation temperature (T_sr [°C]) of the calves (CT_sr). The infrared camera will be placed on a tripod behind the participant at the height of 85 cm. The distance between the camera and participant will be 2.30 m. The focus will be set perpendicular to the backside of the calves to capture the area from the popliteal space to the ankle. To stabilize the running position on the treadmill, a horizontal barrier made of elastic-plastic will be fixed behind the participant, and the foot position will be marked on the treadmill with adhesive tape. The infrared camera temperature scale will be set from 25 °C to 35 °C. An acclimatization period of 10 minutes will be implemented before the exercise test. The participants will only be allowed to wear ankle socks and shorts; should sleep and fast for 8 hours; should drink not more than 1.5 L of water; should be physically inactive; and should avoid cosmetics, showering, and sunbathing for at least 8 hours before the test.

We will perform 3 steps for thermogram analysis: (1) thermogram selection, (2) thermogram processing, and (3) analysis of the region of interest. We will process the image by coloring everything in black, except for the relevant area in the thermogram. This analysis procedure is in line with that of Ludwig et al [108]. The final regions of interests analysis will be performed with the software Analysis of thermal images–TDM V2.0 (Optoprecision GmbH). The report of the entire infrared thermography procedure is in accordance with that of Moreira et al [109].

Before, immediately after, and 30 minutes and 60 minutes after the exercise testing, the median cubital vein will be punctured for venous blood sampling. In total, 100 mL of whole blood will be taken and further processed for the preparation of blood serum and plasma aliquots, depending on the subsequent analysis. Blood counts and physiological parameters will be assessed from the venous plasma or serum, respectively, at the defined time points. The following parameters will be included: creatine kinase, lactate dehydrogenase, hydroxybutyrate dehydrogenase, calcium, cholesterol, triglycerides, high-density lipoprotein-associated cholesterol, low-density lipoprotein-associated cholesterol, total leukocytes, erythrocytes, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin, platelets, neutrophils, eosinophils, basophils, monocytes, lymphocytes, and total protein. Blood plasma samples for molecular marker analysis will be aliquoted and stored at –80 °C before further processing. Venous blood will be used to determine and analyze venous cell-free DNA (concentration using quantitative real-time polymerase chain reaction [110,111]); non–disease-specific epigenetic patterns and sequence analysis using targeted next-generation sequencing [112]; endocannabinoids, catecholamines, and inflammatory lipids using mass spectrometry [113,114]; and further circulating signal transducers (extracellular vesicles and cytokines using multiplexed assays [115-117]). In parallel with venous blood sampling and additionally at each increment of the exercise test, capillary blood samples will be taken from the fingertip to determine capillary cell-free DNA concentrations.

On the basis of lactate kinetics during the incremental stepwise running tests, the individual anaerobic threshold (IAT) will be used to determine the intensity ranges (regeneration, light, moderate, and vigorous) for the training sessions. The IAT will be determined by means of minimum lactate equivalent +1.5 mmol/L [118,119]. Here, we will define 4 training intensities (regeneration: 50%-70%, light: 70%-85%, moderate: 85%-100%, and vigorous: 100%-110%), which are relatively related to the velocity at IAT (100%).

After the baseline examination (T0 and T1), an 8-week sports intervention will be carried out for the IG. The CG will not participate in any training interventions during this period. The training contents are interval- and continuous endurance–oriented. The external load can be individually controlled and adjusted with the help of the IAT and the related relative training intensities. The intervention is divided into 3 mesocycles and 8 microcycles (=1 week): adaptation (2 weeks), frequency (3 weeks), and intensity (3 weeks; Figure 3). One box (Training details) corresponds to a fixed training week from Monday to Sunday. 5 minutes warm up and 5 minutes cool down should be included in every training session. The numbers 1x and 2x are the weekly frequencies of the training session.

Here, a comprehensible protocol is used to adjust the training load using the principle of periodizing and cyclizing (frequency, intensity, time, type, volume, pattern, and progression; Figure 4) [43,45,59]. The gradual progression of the training load is based on weekly feedback of the participant via the Foster scale (0-10) [120]. The value for ailments has a higher priority than the load value. An optimal training load increase over 8 weeks of the training intervention is shown in Figure 3. After 8 weeks, each participant in the IG can end the training intervention on another stage based on the weekly feedback. In addition to the exercise training, the participants can conduct relaxation exercises in the form of stretching and mobilization exercises.

IG and CG participants will receive a smartwatch (M430, Polar Electro Oy) to record training times and monitor daily activities (eg, step count and sitting time). In addition, smartwatch data will be used to control for additional physical activity apart from the training intervention in the IG and physical activity, in general, in the CG. Furthermore, it will be used as a mechanism to control for the effect on primary outcomes by only receiving an intervention. The heart rate and distance of the training session will be recorded by the IG using the M430, which enables comparison of the data reported via the weekly feedback with smartwatch-recorded data. In the CG, the smartwatch is only used to record daily activity. To improve adherence of all participants, 15 smartwatches will be raffled among all participants who have completed the study.

The training intervention described above will be accompanied and remotely supervised with the help of a web-based physical exercise training support via a website (Figure 5) [121]. In a personal appointment after the incremental stepwise running test, the participants in the IG will be introduced to the website and will receive comprehensive information about the training. External access to the website is not possible as only study participants can register. Both the introduction and the remote training supervision will be done by 1 person, which enables the coach to supervise several participants in their training process without any time constraints. Although there is no face-to-face supervision, based on the objective training session heart rate data and GPS data as well as the subjective feedback, adequate supervision can be guaranteed remotely. Participants have the possibility to get in direct contact with the coach via a message function, for example, in case of problems or uncertainties. The weekly training schedule is also transmitted in this manner. If a participant misses the feedback, a reminder is sent via email at the beginning of the following week. In a table that participants can download from the website, they will enter various medical and sports values that are therapeutically relevant after each training session. These data will be used for evaluation and for the training adjustment of the following week, as described in Figure 4. A follow-up is planned at 12 weeks (T4) and 24 weeks (T5) after completion of T3, and participants will be asked via email to complete the web-based questionnaires used at T2.

All data will be recorded and stored in the Department of Sports Medicine, Prevention, and Rehabilitation (Mainz, Germany), and access will be restricted to authorized personnel. Confidential information will be stored safely in the trial files and will not be shared with any other third institution or entity except in response to a legal requirement. Electronic data will be stored on password-protected computers in a restricted-access building. All data will be pseudonymized and submitted to the evaluation team for analysis. A data manager from each department is nominated, who will be responsible for data management and processing. The data and safety management team will be formed by the members of the project. Only the data manager will have access to all personal data. Trial documentation and data will be archived for 10 years after completion of the trial. All dropouts and the reason for dropping out of the study will be reported. Any harm or unintended events during the examinations will be recorded and reported to the Landesärztekammer (State Medical Association) Rhineland-Palatinate. The results will be publicly available in open-access journals and presented via contributions at conferences and congresses according to good scientific practice.

The power calculation was based on the analyses of the primary outcomes, that is, aerobic capacity (VO₂peak) and experimental stress reactivity (saliva cortisol). On the basis of the results of Bacon et al [122], the sample size was determined a priori via power analysis (G*Power V5), yielding a minimum sample size of 67 for a medium-to-large effect size of f(V)=0.35. An alpha error set to .05 and statistical power of 0.80 were assumed for a multivariate analysis of variance (MANOVA) with 2 repeated measurements, 2 groups, and interaction effects. Descriptive statistics will be used for all outcome measures at each time point to give an overview of all results. First, MANOVA will be performed with the main primary outcomes as dependent variables, that is, aerobic capacity (VO₂peak) and experimental stress reactivity (saliva cortisol) to maximize the power of the analyses and to be able to investigate the main and interaction effects of time (baseline [T0, T1] vs post [T2, T3]) and group (IG vs CG). Groups should be equal at preintervention (T0 and T1). A time effect should be observed for the IG. Furthermore, we expect a significant time×group interaction as the IG should react differently over time, with a higher increase in aerobic capacity (VO₂peak). Significant effects of the MANOVA will be followed up by posthoc analyses of variance. As a secondary aim of our study, we aim to identify variables (eg, emotion regulation and biological metrics) that mediate the relationship between physical exercise and resilience outcomes. To that end, we plan to calculate multiple regression analyses that include potential mediators. Principal component analyses regarding potential mediating variables will precede the mediator analyses to avoid redundancy in our regression models. This will be statistically detailed with posthoc tests and corrections for multiple comparisons (Bonferroni-Holm correction [123]). Missing data will be omitted from the data set.

Evidence that exercise can lead to various benefits in mental health and resilience is increasing [16,124-129]. This positive impact was achieved in supervised as well as unsupervised exercise programs. Here, we implement a study design that addresses the link between aerobic capacity and resilience from a multifactorial perspective to understand why some people are more resilient to stress than others. To follow this question, we will investigate the association between innate and trained aerobic capacities in the form of VO₂peak of healthy, physically inactive adults and their resilience expressed primarily by salivary cortisol level changes and emotion regulation ability in response to different stress paradigms. We hypothesize that (1) participants with a high innate aerobic capacity will show a higher resilience to stress compared with participants with low aerobic capacity. Furthermore, we hypothesize that (2) 8 weeks of individually structured interval-type and continuous endurance exercise training remotely supervised by a web-based approach will improve the aerobic capacity of healthy inactive adults and that (3) the change from sedentary to an active lifestyle will improve resilience. In this study setting, we will be able to exploratively examine the molecular link between the effects of acute and chronic physical exercise and improved resilience to stress.

The strengths of our web-based, supervised physical exercise training approach with a focus on increasing VO₂peak in healthy, physically inactive adults are as follows: (1) participants are not location dependent for the training sessions; (2) individualized, gradual, and timely adjustment of training content based on the individual’s feedback reduces the risk of injury, avoids overtraining, and avoids ineffective training; and (3) only 1 sports therapist or trainer is needed for training management of several training groups or trainees, which obviously has economic advantages and reduces bias from different sports therapists. However, web-based training support also has disadvantages compared with presence (face-to-face) interventions. Notably, it requires access to the internet, which limits the study population. Furthermore, we are dependent on the correct and complete documentation of the training sessions by the participants. For an objective way to control reports on the training and to remotely supervise the participants in the IG, we included a smartwatch (Polar M430) that records the heart rate and distance of the individual training sessions. These are given to both the IG and CG participants to exclude an effect on the study outcome only by wearing the smartwatch. Furthermore, group or partner training can bring motivational and social benefits [130-132], which is not directly provided in our web-based physical exercise training support. This may reduce adherence of the participants to the study and, additionally, exclude the social aspect of the effect of an active lifestyle on resilience. Therefore, communication via the web-based platform is intended to be frequent and without any restraints toward the trainer. Finally, inexperience, fear, and the risk of injury can occur during an unaccustomed and intensive training load, which could be reduced or not arise at all through presence support. To counteract this, we have implemented a 2-week adaptation phase in our training intervention that will offer an easy training start and reduce the indicated risks. In addition, we address the individual needs of the participants based on their weekly feedback, so overload is unlikely to occur during the intervention. Adhering to the described improvements in web-based physical exercise interventions will offer the most individual and effective training support with a simultaneously reduced workload for the coach. With this, our research group has already successfully conducted similar studies using the internet for training interventions in patients with Barrett carcinoma, nonalcoholic fatty liver disease, depression, cystic fibrosis, and systemic lupus erythematosus [40-45].

Here, we selected an exercise intervention period of 8 weeks using a combination of interval-type training and continuous endurance exercise to efficiently enhance aerobic capacity in a short period. Interval training improves VO₂peak more strongly in comparison with continuous exercise training [8]. Moreover, psychological benefits in attention, complex attention, and executive function were shown in high-intensity interval training compared with those in low-intensity training [133-135]. However, to counteract overtraining and risk of injury and to promote adherence, we complement interval-type training with endurance exercise sessions. A number of high-intensity interval training and moderate-intensity intervention studies with physically inactive participants decided for a comparable exercise intervention time leading to significant increases in VO₂peak of 3%-19% [9-11]. Although a longer intervention period generally leads to higher increases in maximal oxygen uptake [122], it has been reported that enjoyment of interval training decreases with time [10]. Therefore, we considered a study design that used mixed model exercise sessions with a short intervention time of 8 weeks as efficient to enhance aerobic capacity in physically inactive adults and as suitable to analyze the impact on resilience.

Repeated stress induction and the estimation of resilience factors in a test-retest setting, as chosen in our study, are obstacles to resilience research. It was shown that for psychological stress tests, a test-retest design could lead to a learning effect, resulting in lower stress response [136,137]. Here, we use 2 different stress tests for T0 and T2 to reduce habituation to the applied stressor. By randomly assigning participants to the 2 groups, we intend to obtain the best estimate of the intervention effect on stress responses. However, differences in stress parameters at T2 might not exclusively be attributed to the intervention, as possible placebo effects (ie, simply receiving an intervention) or mediators may need to be considered. Previous studies analyzed similar between-subject comparisons, with less stress reactivity during the Trier Social Stress Test in trained men compared with untrained men [138]. In addition, we account for preintervention differences in stress reactivity and recovery by applying the SECPT at T0. However, as measures from both stress tests are not directly comparable, the experimental design does not permit within-subject analyses. To assess differences in individual changes within the IG, future studies may construct and apply parallel versions of a stress test pre and postintervention.

Several psychological interventions have been similarly used to increase resilience [139], and it might have been interesting to have a sport exercise group; a psychological exercise group that uses a psychoeducational element, such as cognitive restructuring; a mixed model group (sport and psychological); and a CG to obtain the influence of different types of interventions and their combinations on resilience to stress exposure. In our study, we focus on the physiological aspects of increasing VO₂peak and its effect on resilience only. However, the identified relationships between molecular markers and exercise-induced resilience parameters in our setting may be further elucidated in mixed model interventions in follow-up studies.

We cannot exclude a selection bias so that persons who are generally interested in (1) psychological questions, (2) physical exercise, or (3) web-based services are rather addressed by the study announcement. Furthermore, because of a high time commitment for presence tests (2 measurement time points before and after the intervention), (4) people who live close to the survey site are more likely to participate and complete the study. On the basis of the World Health Organization guidelines on physical activity and sedentary behavior, we included individuals who were physically active for <150 minutes/week [140]. These individuals are described as insufficiently active, although this definition is very strict. Possibly, a cutoff point of physical activity <90 minutes/week would obtain a greater effect on the primary outcomes. Here, we decided in favor of the adherence to the study considering the training times that had to be completed.

Resilience is a dynamic and variable state that can be represented in a multifactorial manner. So far, individual instruments have failed to fully capture the variance of resilience [17,141,142]. For this reason, our study uses a wide variety of methods (physiological [blood markers, saliva markers, and hair] and psychological [questionnaires, stress tests, and emotion regulation]) to have a large number of independent predictors to clarify the variance in resilience to the greatest degree of accuracy. Owing to the high number of methods and the collected variables, spurious correlations or pseudoassociations may occur [143], especially when exploratory analysis is performed beyond hypothesis testing. The validity of these findings may be estimated in future studies.

This study will allow us to investigate whether an increase in physical activity with a simultaneous improvement in aerobic capacity is associated with an increase in resilience and whether this effect is reflected in circulating molecular marker levels. These findings will help to reveal novel characteristics of human resilience and may offer novel approaches for the prevention and therapy of mental disorders by exercise prescription.