Author(s): LEPLEY, A., MELVIN, A., AUDET, A., WASSERMAN, M., SOLDES, K., DEJONG LEMPKE, A., KOZLOFF, K., Institution: UNIVERSITY OF MICHIGAN, Country: UNITED STATES, Abstract-ID: 553

INTRODUCTION:
Smart watch technology is widely used by athletes to obtain metrics on fitness and enhance the overall exercise experience. The capability to accurately measure and report on running distance, both indoor and outdoor, is pivotal for providing users with valuable feedback during and post-activity, and can influce pace metrics and workout trends. As individuals increasingly rely on devices to track and monitor workouts, it becomes imperative to evaluate the precision of distance measurements so consumers can make informed decisions. The purpose of this study was to evaluate the accuracy of consumer smart watches in measuring run distance in indoor and outdoor environments.
METHODS:
102 participants (age: 39.7±13.9yrs; BMI: 23.8±4.0kg/m2) completed 2 running sessions on separate days at self-selected pace: 1. outdoor 5km run on the same predetermined GPS measured route, and 2. an indoor 5km run on a treadmill (4Front, Woodway). A subset of participants performed a 3rd trial of either a standardized GPS measured outdoor (n=13) or indoor (n=12) 20km run. During all runs, participants wore a Garmin Forerunner 945 on their right wrist, and an Apple Watch Series 7 (proximal) and Samsung Galaxy Watch 5 (distal) on their left wrist. Distance was recorded from each device after the run and mean absolute error (MAE) and mean absolute percent error (MAPE) were calculated from the known distance from the measured route/treadmill reading. Intraclass correlation coefficients (ICC) with absolute agreement were used to assess agreement between devices.
RESULTS:
Samsung (MAE: 0.10±0.18km; MAPE: 3.2±10.5%), Garmin (MAE: 0.07±0.11km; MAPE: 1.4±2.3%) and Apple (MAE: 0.07±0.19km; MAPE: 1.5±3.8%) watches demonstrated good agreement with each other (ICC(3,3): 0.84, p<0.001) during the outdoor 5km trial and demonstrated less error compared to the indoor 5km trial (Samsung MAE: 0.77±0.46km; MAPE: 15.4±9.3%; Garmin MAE: 0.69±0.71km; MAPE: 13.9±14.2%; Apple MAE: 0.49±0.51km; MAPE: 9.9±10.2%). The devices demonstrated moderate agreement during the indoor 5km run (ICC(3,3): 0.59, p<0.001). Similarly, Samsung (MAE: 0.18±0.18km; MAPE: 0.9±0.9%), Garmin (MAE: 0.05±0.04km; MAPE: 0.2±0.2%), and Apple (MAE: 0.19±0.81km; MAPE: 0.9±0.4%) watches demonstrated moderate agreement (ICC (3,3): 0.52, p=0.04) and less error during the outdoor 20km trial compared to the indoor 20km trial (Samsung MAE: 3.51±1.89km; MAPE: 17.5±9.4%; Garmin MAE: 2.12±1.37km; MAPE: 10.6±6.8%; Apple MAE: 0.93±0.74km; MAPE: 4.6±3.7%). The devices demonstrated poor agreement during the indoor 20km run (ICC (3,3): 0.33, p=0.05).
CONCLUSION:
Consumer smart watches demonstrate less error on distance metrics during outdoor compared to indoor runs. The three watches tested demonstrate moderate-to-good agreement on outdoor distance, and poor-to-moderate agreement on indoor distance measurements. Consumers can use this information when understanding the accuracy of distance metrics measured by smart watch technology in varying environments.