Locomotor-respiratory coupling (LRC) is potentially advantageous for runners but may be difficult to perform. We introduce a custom solution implemented on a smartphone to individualize and guide runners toward LRC.
While running is amongst the most popular activities for competition and leisure, an estimated 20-40% of runners may suffer from respiratory limitations. Some of these runners may benefit from breathing techniques to improve performance or alleviate respiratory discomfort. One such technique is locomotor-respiratory coupling (LRC), a frequency and phase synchronization of breath to step. Studies have demonstrated that LRC may benefit ventilatory efficiency via “step-driven flows,” and some experts have argued it could be used for pacing exercise or increasing positive emotional states. Nevertheless, it may be difficult to perform without coaching or guidance. Here we propose RunRhythm, a custom smartphone application to deliver step-synchronized sound guidance for LRC. This concept builds on previous evidence that sound guidance can be effective and integrates features to maximize adherence and individualization. Preliminary results show that this application is a promising and efficacious method suitable for research on LRC in field exercise. Recommendations for use and further development are discussed to further develop this concept for the benefit of a wider population.
Running is perhaps the most widely popular form of exercise in part due to its accessibility and range of physical and mental health benefits1,2. Nonetheless, many aspiring runners struggle to start or maintain running habits. This could be due to breathing difficulties, which affect an estimated 20-40% of runners3,4,5. Reducing exercise-induced breathlessness is theoretically possible with the use of specific breathing techniques, but the exact methods, risks, and benefits of doing so are unclear. While improving fitness and/or slow-paced breathing at rest may alleviate respiratory discomfort during exercise6,7, these solutions take weeks or months to realize benefits. Some research has speculated that the direct implementation of breathing techniques during exercise may be more effective at producing acute benefits8, although such investigations are limited. Digital tools that enable individualized instruction might be needed to perform such studies in an effective intervention format.
Locomotor-respiratory coupling (LRC) is a synchronization phenomenon in which breathing and movement are frequency- and/or phase-synchronized. In specific exercises such as running, LRC indicates a whole-integer ratio between breathing (BR) and step rate (SR), as well as phase locking of footstrike to breath onset (i.e., stepping exactly on inspiration). LRC can be expressed volitionally or automatically and may emerge as a learned behavior with exercise training9. Humans naturally synchronize gait with interfering auditory noises (including breathing) perhaps to reduce auditory stimulation, which suggests LRC has evolutionary phenomenology10. Some reports indicate that LRC benefits movement economy and performance, and reduces breathlessness11,12,13,14,15. Some studies reported negligible benefits16,17,18. Any physiological benefits could be related to "step-driven flows": each footstrike creates downward momentum of the abdominal contents (the "visceral piston"), which when synchronized with the onset of inhale or exhale can be additive to ventilation.
Daley et al.19 measured ventilatory flow and impact forces during treadmill running and speculated that step-driven flows can contribute up to 10-12% to total ventilation. They also reported that it could speed up ventilatory transitions. Other neuromechanical mechanisms may interact with this phenomenon9. While the visceral piston is a result of precise phase coupling, frequency coupling may be independently valuable, especially to the novice runner. BR is closely related to effort across different exercise intensities20. As SR is generally stable and associated with running speed21, LRC may support self-awareness and enable easy pacing of BR and, thus, intensity throughout the run. Finally, LRC at uneven ratios (e.g., 5:1 steps per breath) could help to prevent exercise-related transient abdominal pain ("side stitch"). A majority of runners experience this temporary, but distracting and painful symptom each year22, often leading to their need to stop running. One theory of side stitch etiology is that repeated breathing on the same side footstrike may irritate the phrenic nerve. Thus, it might be avoided by LRC at uneven ratios, which leads to breathing on alternate legs.
Few reports have discussed how to support runners in performing LRC. At least two studies have exhibited biofeedback-style methods14,23 while many utilized simple verbal coaching24,25. While these methods have shown promise in stimulating LRC acutely, they are highly standardized and require specialized equipment. As such, they are likely not suitable for field applications, nor are they accessible to most runners. Regardless, sound guidance is a natural choice since humans intuitively synchronize movement to predictable auditory events (metronome or music)26. Applications should thus carefully consider sound tempo and structure in the context of motor learning. While simple, constant-tempo audio is predictable and efficacious for stimulating entrainment, it contradicts the naturally nonlinear behavior of step and respiratory rhythms in healthy runners27,28. Changing a runner's preferred SR might reduce running economy 29 or could modify injury risk factors30. Thus, sound instructions should be continuously adapted in real time to follow the runner's SR31.
We recently introduced a concept that integrates the above recommendations into a simple, user-friendly, custom smartphone application32. The first iteration allows for the selection of a single LRC ratio to be instructed throughout the run. The phone's stock SR algorithm is leveraged to provide real-time SR information to the application. Then, step-synchronized sounds are produced indicating when the runner should exhale and inhale: a high-pitched tone for steps during inhalation, and a low-pitched tone during expiration. Prescribed LRC ratios were derived from a control visit without breathing instruction. We found a large increase in LRC from 26.3 ± 10.7% to 69.9 ± 20.0% of the run with the application instruction during outdoor submaximal running. Limitations noted with the protocol and application include extensive familiarization required, limited sample size, and constant sound instruction. Hence, a new version of this application was developed to improve user experience and enable more broad testing and experimentation in field exercise. This application is titled RunRhythm since its intended purpose is to support runners in finding and maintaining a rhythm during running. It will be referred to hereafter as the app.
The purpose of this report is to introduce a new digital tool and methodological approach that enables intuitive and field-ready LRC guidance for research studies involving experienced or aspiring runners. The app is a research-grade application in beta testing for Android devices. The core functionalities of the application are SR detection and LRC guidance. When running is detected, breathing sounds are created according to the selected settings in the user interface. The application calculates SR from the phone accelerometer using one of two algorithms: either the factory SR algorithm implemented by the device manufacturer or a custom SR algorithm created by the application manufacturer. Both algorithms produce a constant livestream of SR, which is then smoothed on a moving average according to an adaptive window. The window size is dynamic to balance reactivity and outlier smoothing. The result is a constantly updated value of live SR.
Since the app calculates SR from the device movement, the placement of the phone on the body is of utmost importance. Most stock SR algorithms are position-agnostic, and therefore, can be placed on any part of the body during running to produce accurate SR values. The custom algorithm implemented here also behaves as such. However, a firm placement closer to the center of mass may improve the stability of the SR detection, and as a result, the sound quality produced by the app. Pilot testing shows that placements with 1-dimensional oscillation (i.e., vertically up and down such as in a chest pocket or waist pack) may perform better than those with 2-dimensional movement (i.e., swinging such as in a thigh pocket or armband).
SR data are fed to an integrated sound engine (see the Table of Materials). Step sounds are played only if the system detects SR > 0. When the SR is above a preset threshold (determined in backend settings [protocol section 3.6]; i.e., 120), the application understands that the user is running and triggers the start of breathing guidance sounds. Then, this live SR value is used to set the tempo of the step and breathing guidance sounds as long as a "running" SR value is maintained. When SR > threshold, the sounds generated match the tempo of SR by default. The exception is when the backend setting "sound tempo" is altered (determined in backend settings [protocol section 3.5]). For example, with a selected upper limit of 180, even if the runner begins running at a higher SR of 185, the sound tempo will not exceed 180. When they lower their SR down to 175, the sounds will lower to 175, continuously adjusting within the preset limits. As described in protocol step 3.5, these sliders allow the user or researcher to set limits on the minimum and maximum sound tempo (bpm). The app allows different LRC ratios (steps:breath) to be selected before or changed during the run. The number of steps per breath phase can be changed from 2 to 9; i.e., a 2:3 ratio reflects 2 steps per inhale and 3 steps per exhale.
Different "soundscapes" were designed to provide a pleasant audio experience to more runners with diverse music tastes based on user feedback and early in-lab experiments33. They have different sounds mapped to the real-time step rate, the instructed breathing phases, and background ambient noise. Step sounds are simple beats that play at the tempo of each footstrike (i.e., right and left steps). Breath sounds integrate several sonic elements and play at a much slower tempo depending on the chosen LRC ratio. The soundscapes available are tribal: organic and instrumental with sharp breathing transitions and step sounds; calming: light and ocean-inspired with smooth transitions and step sounds; energizing: electronic and driving with sharp transitions and step sounds; minimal: simple and smooth with only breath sounds (no step sounds).
The voiceover feature adds simple voice cues corresponding to best-practice research findings regarding LRC familiarization. It provides a series of instructions at the start of the run and then every 5 min after that. First, it states the selected LRC ratio. Then, it states the intended breathing phase in sync with the sound cues for the first three breath cycles. Then it reminds the user, "find your step rate, and step to the beat." For each run, a pre and post run questionnaire is integrated to add subjective feeling data to each run. The subjective vitality short scale34 asks a single item regarding the runner's feeling from 0 to 10. A 0-10 rating of the fatigue scale asks the user to rate their current state of fatigue. Finally, a 0-10 scale rates the degree of breathlessness currently experienced. All of these scales are asked before and after each run. Only after the run, the user is asked to rate their experience of the run's intensity (i.e., light, medium, high, intervals). Users can change the LRC ratio and temporality during the run using the on-screen interface or headphone controls. This may help users feel agency during the run and enables the exploration of personal suitability. In addition, the ratio may need to be quickly changed in response to running events (e.g., hills, fatigue). This protocol includes a description of how to run with the app and later recommendations for its use within research protocols of various types (i.e., indoor, outdoor, interventional, cross-sectional).
This study was granted ethical approval by the Ethics Committee of the University of Salzburg (reference number: GZ 13/2021), and the participants gave their informed consent.
1. Getting started with RunRhythm
Figure 1: App tutorial. RunRhythm provides an introductory tutorial upon its first opening, including details regarding locomotor-respiratory coupling, an animation showing how the application works, and tips for use. Please click here to view a larger version of this figure.
2. Basic functionality
Figure 2: Main app interface. (A) Locomotor-respiratory coupling ratio can be changed from the user interface with spinners that range from 2 to 9. Each value represents the number of steps per breath phase; i.e., 2:3 represents 2 steps per inhale: 3 steps per exhale. The lock icon can be used to fix the ratio difference; i.e., when locked at 2:3, shifting "up" changes the ratio to 3:4 (maintaining a difference of 1 step more per exhale). (B) Soundscape selection allows the user to choose from four predetermined sound layers: tribal, calming, electronic, and minimal. (C) Temporality toggle allows the user to choose from three predetermined settings for guidance frequency: full, medium, and off. (D) Voiceover toggle allows the user to turn voice cues on or off. Please click here to view a larger version of this figure.
Figure 3: Pre and post questionnaires. Identical questionnaires are presented at the start and end of every run. They must be answered to start or finish the run. Please click here to view a larger version of this figure.
3. Backend settings
NOTE: Key parameters affecting application functionality can be changed by tapping on the three dots in the top right corner of the main interface. The default values reflect the recommended values but can be changed. This screen (Figure 4) contains the following settings:
Figure 4: Backend settings. Backend settings include auto pause, a toggle for step detection, and identifier codes. (A) Sound tempo threshold settings allow precise selection of step rate thresholds that bound generated sound guidance tempo. For example, selecting a lower threshold of 155 and an upper threshold of 180 ensures that the sound guidance will not deviate from the interval [155, 180], regardless of the actual SR detected. The default is [0, 200]. Please click here to view a larger version of this figure.
4. Running with the app
Figure 5: In and post-run interface. (A) During the run, a simplified interface is available that allows the user to change key parameters including locomotor-respiratory coupling ratio and temporality. It also displays the current running pace and step rate. (B) After the run, a summary screen displays key metrics including total distance, average pace, and average step rate. Please click here to view a larger version of this figure.
5. Research with the app
NOTE: The above functionalities were developed to maximize user experience and enable research studies of LRC across diverse contexts. The steps given below describe how to integrate the app into a study from initial familiarization to post run application logs. These study methods have been reviewed and approved for experimentation with humans by the ethics committee of Paris Lodron Universität Salzburg (EK-GZ 29/2023).
Figure 6: Researcher familiarization. In research contexts, familiarization by the primary investigator is recommended to ensure a conceptual understanding of locomotor-respiratory coupling and proper use of the application. Please click here to view a larger version of this figure.
The app is the second iteration of this application designed for the purpose of supporting LRC and delivering an audio breathing guidance experience. Numerous pilot studies and one journal publication have been performed supporting its efficacy and confirming positive user experience. In a cross-sectional study investigating the acute effects of LRC instruction (mentioned in the Introduction), it was found that running with guidance greatly increased LRC in 17 novice runners32. For example, one participant increased LRC "attachment" (a quantification of adherence to whole-integer LRC ratios) from 7.9% to 84.1% (figure) at a ratio of 2:3 steps per inhale:exhale (5:1 steps per breath) (Figure 7). The overall group change in entrainment increased from 26.3 ± 10.7% (control) to 69.9 ± 20.0% (with LRC guidance).
Figure 7: Example coupling data. Example of LRC prevalence in one participant during a 30 min easy run (A) without and (B) with app guidance. The left y-axis (blue) displays LRC ratio (steps per breath), and highlighted sections in black correspond to a period of LRC "attachment" > 30 s. The right y-axis displays the discrete relative phase of steps within the breath cycle, where 0° corresponds to inspiration. During the attachment periods, note the parallel lines in the discrete relative phase reflect consistent phase coupling of breath to step. (A) In the control run without guidance, note there was only a brief period (~8% attached) of LRC to a ratio of 2:2 (4 steps per breath). (B) In the run with app guidance, a ratio of 2:3 (5 steps per breath) was instructed; note the substantial, long periods of LRC to the instructed ratio (~84% attached). Abbreviation: LRC = locomotor-respiratory coupling. Please click here to view a larger version of this figure.
In a pilot intervention study, four novice runners ran for 4 weeks with the application (total 10 runs) and instructions to test various LRC ratios. They completed pre and post tests indoors (incremental treadmill test to exhaustion) and outdoors (self-paced 30 min run) to determine how their LRC and other running characteristics changed with the intervention and application use. More specifically, the pre and post tests were completed without LRC guidance or instruction to observe the natural, habituated LRC behavior before and after the intervention. We found marked differences in LRC behavior after the intervention, including not only an increase in overall LRC but also utilization of various ratios and a general reduction in exercise BR. For example, one participant increased LRC from 39.7% to 53.4% after the intervention (Figure 8). Notably, while in the pretest, they performed LRC exclusively at ratios of 2:2 and 2:1 near maximum intensity, in the posttest they used LRC ratios 3:4 and 3:3 early and throughout the middle of the test. Not only were they coupling more, but they were utilizing different ratios at different times, resulting in lower average BR and also increased ventilatory efficiency (VE/VCO2 slope). Other participants had similar changes in LRC (+40% outdoors, +20% indoors), and the one participant in the control condition (no LRC guidance) saw no change (<25% overall, considered nonsignificant). There were some individual trends in subjective vitality response to exercise pre vs. post-intervention showing favorable increases in vitality with LRC (Figure 9).
Figure 8: Example coupling data 2. Example of LRC prevalence in one participant during a maximal incremental treadmill test to exhaustion (A) before and (B) after a 4 week intervention running with the app. The left y-axis (blue) displays the LRC ratio (steps per breath), and highlighted sections in black correspond to a period of LRC "attachment" > 30 s. The right y-axis displays the discrete relative phase of steps within the breath cycle, where 0° corresponds to inspiration. During the attachment periods, note the parallel lines in the discrete relative phase reflect consistent phase coupling of breath to step. (A) In the pre-test before the intervention, note the two periods (~40% attached) of LRC to a ratio of 2:2 (4 steps per breath) and 2:1 (3 steps per breath) near the end of the test. (B) In the post-test after the intervention, not only were there longer periods of LRC (~53% attached), but longer ratios were used (4:4, 3:4, and 3:3), and more LRC was performed earlier in the test at lower relative exercise intensities. Abbreviation: LRC = locomotor-respiratory coupling. Please click here to view a larger version of this figure.
Figure 9: Individual subjective vitality scores recorded throughout the indoor running test pre and post-intervention. P01 and p03 received LRC guidance, while p04 was in the control group. The other participant (p02) has faulty questionnaire data (not shown in the figure). Note the diverging scores for p01 and p03, where the post test trends show increasing vitality after the run. Abbreviation: LRC = locomotor-respiratory coupling. Please click here to view a larger version of this figure.
This methodology presents one of the first evidence-based, field-ready digital tools for instructing LRC to runners. Early results suggest that it is effective not only in quickly learning and adhering to LRC but can also be taught over time and retained. While LRC may emerge naturally with increased running experience, novices are less likely to perform it37. Coincidentally, they are especially likely to experience barriers to participation that include poor fitness and respiratory limitations38,39. Other populations such as obese individuals have even more potential benefits since they experience increased work of breathing and soft-tissue bouncing40,41,42; this dual problem may be addressed via LRC. LRC may help these individuals to run with less discomfort, and digital tools such as this application are accessible modes to deliver it to a larger number of such runners. It is plausible that LRC, and hence, this app could be additionally valuable in contexts that present additional respiratory challenges such as hypoxia, thoracic load carriage, and ultramarathon.
The app utilizes sound guidance to instruct LRC because most runners listen to audio while running43, rhythmic audio motivates and reduces fatigue during exercise31, and it is minimally invasive to other critical senses (i.e., vision). Laboratory studies on LRC have utilized sound cues with a metronome14, which does not consider the natural variability and complexity of gait dynamics during running. Another study used haptic guidance during field running; they reported low adherence to LRC instruction around 26%44. Considering the relatively higher adherence reported here (Figure 8 and Figure 9), it is suspected that auditory guidance is preferable.
While the app does detect and provide live feedback for SR, the BR guidance to specific LRC ratios is predetermined by the application settings. It neither senses nor adjusts based on the user's actual BR. While an individual calibration can be performed to approximate the appropriate LRC ratio choice (protocol section 7.1), without BR sensing, there is no mechanism to confirm adherence or adjust the instruction accordingly. The addition of a BR sensor would improve the application in these ways. For example, a recent review suggests that continuous feedback may have negative effects on learning and habituation45. Other strategies such as bandwidth (guidance only when the individual is outside of a predefined success threshold) or self-determined feedback may be most effective to support learning and self-efficacy45. Finally, the addition of a BR sensor could also provide feedback on phase coupling. Since this element of LRC is central to the visceral piston mechanism and resultant physiological benefits, it should be considered for future feedback applications.
The LRC individualization is essential to effective research applications. Performing this individualization ensures that LRC instruction does not substantially change the runner's BR from their natural one. Alternatively, if changing BR is desired, it enables the user to make changes more sensitive to their typical physiological state. This individual calibration should be done context-specific since BR is trait and state-dependent. For example, BR is typically elevated at higher relative exercise intensities20; hence, the natural LRC ratio may be different during long, slow duration running versus high-intensity threshold running. It may also be affected by factors like environment (indoor vs. outdoor, cold vs. hot), time of day, and biological sex46. Performing this calibration requires measurement systems that can estimate BR and SR synchronously. This can be done with standard sports science equipment such as spiroergometry and accelerometers. In the field, wearable sensors have demonstrated their potential to measure these parameters accurately36. BR and SR can be averaged over a given period to estimate the approximate LRC ratio. When appropriate, data may be trimmed to specific phases within the run since physiological demands (thus BR) and speed (thus SR) may vary. Consequently, the "natural" LRC ratio may vary within a run.
Specific elements of this application should be updated to improve user interaction. While some parameters can be changed with headphone controls, others (such as soundscape) cannot be. One might consider integrating gestures or smartwatch integration for easier interaction. In addition, the next iterations might consider some form of instruction "reset" to use when the runner loses their rhythm. Natural breathing interruptions such as coughing or swallowing, or simple distraction, might cause a runner to become unsynchronized with the guidance. Instead of holding the breath to wait for the next LRC cycle to come, a button, voice command, or sensor-detected event could be used to resynchronize the sound to the runner. Finally, since a majority of runners listen to music43, the app may detract from runners' experience since they cannot listen to their favorite music and receive LRC guidance on "full." Integrating LRC guidance into music itself might allow for an intuitive and natural experience to deliver breathing guidance and enjoy one's preferred music. This is not yet integrated into the application.
Some limitations to this method have already been identified in testing thus far. First, it is possible that the step sounds pull the runner away from their typical SR. While the step sonification only reflects the runner's real-time SR, the moving averaging implies that the sonified SR is initially lower than the runner's immediate SR. Thus, it may be valuable for the user to ignore the step sounds until they have established their own preferred SR for some seconds. Second, related to this previous point the step sounds are automatically played at the user's current SR, but they are not necessarily phase-synchronized with footstrike. Hence, it is possible that the user's SR matches the sonified SR but in antiphase coordination. This may result in undesirable LRC dynamics since the user may lose phase-synchronization of step and breath. This can be corrected with researcher/expert guidance observing the user with the audible guidance sounds. Next, the user selection of LRC ratios is desired for flexibility to diverse runners' needs; however, it could lead to the accidental selection of undesirable ratios. For example, 2:2 may be an appropriate ratio for moderate and high-intensity running but may lead to hyperventilation at lower relative intensities. Highly unequal ratios (e.g., 3:6) may be useful for cueing long exhales, while the reverse (6:3) is undesirable for its excessively short exhale phase47. Finally, it is acknowledged that an application for only the Android platform is inherently limited since many mobile users prefer other platforms. Pilot testing revealed that the application performs as intended only on updated modern mobile devices.
We hope to see this application used in intervention studies of runners who might benefit most greatly from breathing guidance. Novices, females, adolescents, and those who experience exercise-induced breathlessness might experience the greatest benefits from LRC guidance8,48. Experimental methods might be enhanced with the use of the "sound tempo" feature to support desired pacing strategies. While untested thus far, other cyclic endurance activities (e.g., cycling, cross-country skiing) might benefit from LRC guidance such as that provided by the app. More work is needed to assess the suitability of this solution for such applications.
The authors have nothing to disclose.
This work was supported by the Austrian Federal Ministry for Climate Action, Environment, Energy, Mobility, Innovation and Technology under Contract No. 2021$$-$$0.641.557 and the federal state of Salzburg under the research program COMET-Competence Centers for Excellent Technologies-in the project Digital Motion in Sports, Fitness and Well-being (DiMo; Contract No. 872574).
Android smartphone | Samsung or Google | Minimum Android 8.0 required for application functionality | |
FMOD engine | Firelight Technologies Pty Ltd | Sound engine | |
Hexoskin smart shirt | Carré Technologies | Wearable sensor shirt | |
RunRhythm application for Android | adidas GmbH and abios GmbH |