Using GPS, accelerometry and heart rate to predict outdoor graded walking energy expenditure

To determine the best method and combination of methods among global positioning system (GPS), accelerometry, and heart rate (HR) for estimating energy expenditure (EE) during level and graded outdoor walking.

Thirty adults completed 6-min outdoor walks at speeds of 2.0, 3.5, and 5.0kmh^-1 during three randomized outdoor walking sessions: one level walking session and two graded (uphill and downhill) walking sessions on a 3.4% and a 10.4% grade. EE was measured using a portable metabolic system (K4b²). Participants wore a GlobalSat^® DG100 GPS receiver, an ActiGraph™ wGT3X+ accelerometer, and a Polar^® HR monitor. Linear mixed models (LMMs) were tested for EE predictions based on GPS speed and grade, accelerometer counts or HR-related parameters (alone and combined). Root-mean-square error (RMSE) was used to determine the accuracy of the models. Published speed/grade-, count-, and HR-based equations were also cross-validated.

According to the LMMs, GPS was as accurate as accelerometry (RMSE=0.89-0.90kcalmin^-1) and more accurate than HR (RMSE=1.20kcalmin^-1) for estimating EE during level walking; GPS was the most accurate method for estimating EE during both level and uphill (RMSE=1.34kcalmin^-1)/downhill (RMSE=0.84kcalmin^-1) walking; combining methods did not increase the accuracy reached using GPS (or accelerometry for level walking). The cross-validation results were in accordance with the LMMs, except for downhill walking.

Our study provides useful information regarding the best method(s) for estimating EE with appropriate equations during level and graded outdoor walking.