Instrumenting gait assessment using the Kinect in people living with stroke: reliability and association with balance tests

Table 5 Results for the clinically-derived and Kinect-derived multiple regression models

Functional assessment	Clinical			Kinect
Functional assessment	R ²	Adj R ²	AIC	R ²	Adj R ²	AIC
Overall physical function
Timed up and go	0.89	0.88**	81	0.95	0.94**	63
Static balance
Static double leg eyes closed balance	0.41	0.37*	−8	0.47	0.36*	−5
Static weight-bearing asymmetry	0.49	0.44**	102	0.50	0.38*	108
Dynamic balance
Functional reach	0.43	0.39*	109	0.65	0.58**	100
Step test	0.61	0.58**	70	0.83	0.79**	52
Medial-lateral weight shifting ability	0.48	0.44**	60	0.75	0.69**	46

Clinically-derived model included 2 input variables: 10MWT time; total number of steps. Kinect-derived model included 5 input variables: Mean velocity; Foot swing velocity asymmetry; Peak – Mean velocity difference; Step length asymmetry; Affected foot swing velocity.
*p-values <0.01; **p-values <0.001.
R², Ezekiel equation adjusted R² and the Akaike Information Criterion (AIC) score are provided for comparison of model strengths.

ISSN: 1743-0003