Validation Of Vo2max And Sweat Loss Metrics From Wearable Smart Watches

Commercial wearable smart watches offer insights into various physiological metrics, including estimates of maximal oxygen consumption (VO2max) and sweat loss (SL) during activity. However, these measures are generated through proprietary algorithms of uncertain accuracy. It is imperative to validate the physiological outputs from these devices and determine whether they are providing individuals with accurate and reliable data to make informed decisions about their individual health and fitness. PURPOSE: Validate VO2max and SL metrics from consumer grade wearable smart watches. METHODS: 84 participants (42 M/42F; age: 39 ± 13 yrs; BMI: 23.4 ± 3.7 kg/m2) completed three visits: 1. treadmill graded maximal exercise test with metabolic cart to assess VO2max; 2. outdoor GPS measured 5 km run (20.6 ± 4.6oC; 60.1 ± 12.2% humidity; 12.1 ± 6.3 days between visit 1 and 2) at self-selected effort (RPE: 12.9 ± 2.1); and 3. indoor treadmill 5 km run (22.3 ± 1.2oC; 45.4 ± 7.0%; 11.4 ± 8.6 days between visit 2 and 3) at self-selected effort (RPE: 12.7 ± 2.0). Participants wore 2 different smart watch models, one on each wrist, during visits 2 and 3. Nude body mass was measured pre and post 5 km runs to calculate SL (mL). Watch derived VO2max estimates from both devices during the outdoor 5 km run were compared to the in-lab measurement of VO2max using interclass correlation coefficients (ICC) for absolute agreement, mean absolute (MAE) and mean absolute percent error (MAPE). Only Device 1 provided post-run SL estimates, therefore, Device 1-derived SL values from the outdoor and indoor 5 km were compared to actual SL for each trial using ICC, MAE, and MAPE. RESULTS: Device 1 (ICC1,2 = .91; p < .001; MAE: 4.1 ml/kg/min; MAPE: 8.8%) and Device 2 (ICC1,2 = .91; p < .001; MAE: 3.6 ml/kg/min; MAPE: 8.0%) smart watches provided excellent agreement for VO2max estimation during the outdoor 5 km run. Device 1 provided poor agreement for SL during the outdoor (ICC1,2 = .35; p = .01; MAE: 156.5 ml; MAPE: 30.1%) and indoor (ICC1,2 = .15; p = .13; MAE: 164.9 ml; MAPE: 28.9%) 5 km trials. CONCLUSION: VO2max, but not SL, can be accurately estimated using commercial wearable smart watches. Further validation is needed for different activities and environmental conditions. This study received funding from Samsung Electronics Co., Ltd. This study received funding from Samsung Electronics Co., Ltd.