Development and Verification of Postural Control Assessment Using Deep-Learning-Based Pose Estimators: Towards Clinical Applications

Table 1

Regression modeling results.

Model	Predictor	(SE)	value	AIC	Adjusted
Adult and child datasets (combined)
Selected variables with duration time	SPB1	-0.71 (0.14)	<0.001	-112.10	0.86
	SPB3	0.42 (0.12)	<0.001
	AG1	-0.22 (0.12)	0.081
	AG2	-0.15 (0.13)	0.26
	Duration time	0.97 (0.08)	<0.001
Selected variables without duration time	SPB1	-1.47 (0.20)	<0.001	-14.85	0.68
	SPB3	0.60 (0.18)	<0.001
	AG1	-0.28 (0.19)	0.14
	AG2	-0.21 (0.20)	0.30
Duration time only	Duration time	1.41 (0.07)	<0.001	-55.61	0.77
Child dataset only
Selected variables with duration time	SPB1	-0.27 (0.12)	0.032	-111.02	0.84
	SPB3	0.10 (0.09)	0.28
	AG2	-0.23 (0.06)	<0.001
	Duration time	0.82 (0.07)	<0.001
Selected variables without duration time	SPB1	-0.76 (0.21)	<0.001	-28.57	0.45
	SPB3	0.21 (0.17)	0.22
	AG2	-0.30 (0.10)	0.0051
Duration time only	Duration time	1.00 (0.07)	<0.001	-88.07	0.77
Adult dataset only
Selected variables with duration time	SPB1	-1.11 (0.32)	0.0013	-54.73	0.83
	SPB3	1.61 (1.16)	0.17
	AG2	-0.69 (0.18)	<0.001
	Duration time	1.51 (0.28)	<0.001
Selected variables without duration time	SPB1	-1.00 (0.42)	0.021	-31.51	0.71
	SPB3	-2.75 (1.09)	0.016
	AG2	-0.47 (0.22)	0.039
Duration time only	Duration time	1.89 (0.24)	<0.001	-16.60	0.58