**6. Conclusion**

Studies suggest that optimizing cadence is an important factor in reducing the risk of sustaining a running-related injury and in improving overall running performance. In this work, we presented a feasibility study utilizing an mHealth solution to long-distance running cadence-based coaching, called RunningCoach. Future versions of the system will include a music player that selects music with beats that are on the desired cadence for the day. The feedback from the subjects in this study will also be incorporated in the next version of the system.

Based on the findings of the study, there are early signs of satisfaction from a usability and perceived accuracy point of view, with one exception. The video-based heart rate estimates were perceived as inaccurate in this study. As such, the study findings indicate that there is a need for tools that systematically assess the accuracy of sensory estimates and guide the estimation algorithms accordingly. For example, the algorithms for estimating cadence would be different when the phone is secured on the hip versus when the phone is held by the runner in her or his hand. In these cases, it is the responsibility of the system to employ the correct estimation algorithm by detecting the conditions under which the system is being used. More generally, the study findings suggest that audit mechanisms need to be developed and employed for each estimation algorithm, in order to ensure, verify and quantify the accuracy of its outputs.

One important usability issue to be studied in the future is the motivation of runners to use this and similar mHealth technologies for tracking their runs. There is clearly an interest of (novice) runners to improve cadence for performance gains as is evident from a number of apps that provide feedback through music or otherwise [5–10]. However, the evidence for the motivation of using apps for injury prevention is to the best of our knowledge limited. There are several behavioral factors, specific to runners [49], that influence their attitudes towards the level of training and higher risk of injuries. As noted by [50], injury-preventive actions that require behavior modification need to take into account that runners' perceived susceptibility to sport has multiple predictors, including previous experiences, neuroticism and obsessive passion. Mobile applications for runners thus provide an opportunity to address injury prevention through individualized feedback and various motivational mechanisms, which were out of the scope of this pilot study.

**Acknowledgments:** We would like to thank David M. Liebovitz, MD for suggesting long-distance running as an application to the Berkeley Telemonitoring Project and advising on the design of the app. Eugene Song, Uma Balakrishnan, Hannah Sarver, Lucas Serven and Carlos Asuncion contributed to the design of the study protocol, and to the design and implementation of the RunningCoach system. Kaidi Du, Caitlin Gruis, Yu Xiao contributed to the implementation of the RunningCoach system that was used in this study. We would like to thank everyone involved in the Berkeley Telemonitoring Project at UC Berkeley for their hard work that made this work possible. We would also like to thank the anonymous reviewers and the editors for their invaluable suggestions and assistance. This work was supported in part by the Center for Long-Term Cybersecurity (CLTC) at UC Berkeley, including the fees to publish in open access. The views expressed in this paper are those of the authors and do not necessarily represent the official views of the CLTC.

**Author Contributions:** D.A. and R.B. contributed to the design of the study protocol and conducted the study. D.A. and G.X.P. contributed to the design and implementation of the RunningCoach system used in this study. D.A., G.X.P. and G.K. analyzed the data. All authors contributed to the writing of the paper.

**Conflicts of Interest:** The authors declare no conflict of interest. The funding sponsors had no role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript; nor in the decision to publish the results.
