Validity of treadmill- and track-based individual calibration methods for estimating free-living walking speed and VO2 using the Actigraph accelerometer

Table 1 Regression models of prediction error for walking speed and VO₂ by calibration method

	Prediction error for speed of walking (km · h⁻¹)		Prediction error for VO₂ (ml · kg⁻¹ · min⁻¹)
Regression parameter	Treadmill method	Track method	Treadmill method	Track method
	Coefficient (95 % CI ) BPV	Coefficient (95 % CI) BPV	Coefficient (95 % CI) BPV	Coefficient (95 % CI) BPV
Intercept	0.71* (0.56, 0.87) 0.54*	0.02 (−0.05, 0.08) 0.19***	4.99* (3.96, 6.00) 3.26*	0.33 (−0.06, 0.73) 1.10***
Speed	0.15* (0.06, 0.24) 0.26*	−0.04 (−0.09, 0.01) 0.08*	1.16* (0.58, 1.75) 1.57*	−0.29** (−0.49, −0.08) –
(Speed)²	−0.06* (−0.11, −0.01) 0.13*	−0.04 (−0.09, 0.01) 0.10**	– – –	−0.20* (−0.38, −0.02) –

These regression models report on average prediction error by calibration method (see intercept values) and if this error depends on walking speed (see regression coefficients for speed and speed²). Coefficient = regression coefficient; 95 % CI = 95 % confidence interval; BPV = Between-participant variability in regression coefficients (expressed as standard deviations); – = not applicable (not significant regression term or between-participant variability based on a comparison of fixed and random slope models); *p < .05; **p < .01; ***p < .001. Speed (km · h⁻¹) was centered at 5 · 7 km.h⁻¹

ISSN: 2052-1847