TABLE 1.

Outline of the gait event identification methods adopted in this study.

Feature	Marker trajectory	References	Method	Component	Initial contact detection	Final contact detection
Position	mHEEL, mTOE, and mPELVIS	Zeni et al. (2008)	M1	AP	Local maxima of mHEELAP from mPELVIS	Local minima of mTOEAP from mPELVIS
	mHEEL and mTOE	Desailly et al. (2009)	M2	AP	First maximum between high pass filtered mHEELAP and mTOEAP	Last minimum between high pass filtered mHEELAP and mTOEAP
Velocity	Mid-point between mHEEL and mTOE	O’Connor et al. (2007)	M3	V	Local minima	Local maxima
					Additional constraints: timing constraints, and threshold on vertical marker position	Additional constraint: timing constraints
	mHEEL and mTOE	Ghoussayni et al. (2004)	M4	AP and V	Sagittal vHEEL lower than 0.5 m/s a	Sagittal vTOE higher than 0.5 m/s a
		Enhanced M4	M5	3D	Rearfoot contacts: 3D vHEEL lower than 0.5 m/s b	3D vTOE higher than 1.0 m/s and then refined when the vHEEL decreases after its local peak b
					Forefoot contacts: 3D vTOE lower than 0.5 m/s	—
		M4 modified as suggested by Bruening and Ridge (2014)	M6	AP and V	Sagittal vHEEL lower than 0.78 * walking speed c	Sagittal vTOE higher than 0.66 * walking speed c
		Enhanced M6	M7	3D	Rearfoot contacts: 3D vHEEL lower than 0.5 * walking speed b	3D vTOE higher than 0.8 * walking speed c and then refined when the vHEEL decreases after its local peak b
					Forefoot contacts: 3D vHEEL lower than 0.8 * walking speed b , c	—
Acceleration	mHEEL and mTOE	Hreljac and Marshall (2000)	M8	AP and V	Local maxima of aHEELV	Local maxima of aTOEAP
					Additional constraints: relevant jerk is null	Additional constraints: relevant jerk is null
	mHEEL and mTOE	Hsue et al. (2007)	M9	AP and V	Local minima of aHEELAP	Local maxima of aTOEAP
Position and velocity	mHEEL, mTOE, and mPELVIS	Combination of M1 and M7	M10	3D	Events initially detected using M1 and then refined according to the M7 conditions	Events initially detected using M1 and then refined according to the M7 conditions

^aVelocity threshold increased from 0.1 m/s to 0.5 m/s as suggested in Bruening and Ridge ( 2014).

^bVelocity thresholds adapted to the observed 3D velocities.

^cWalking speed calculated for each test as the average stride speed; initial contacts detected with the method M5.