Table 4.

The top 10 ranked gait features and demographic variables in the PLS prediction model and their latent variable number.

Rank	Feature name	Feature type	Direction	Latent var #*
SINGLE-TIME FALL PREDICTION
1	Harmonic Ratio	Gait	AP	1 (7)
2	Number of falls (past 6 month)	Other	–	2
3	Low freq. percentage	Gait	ML	4
4	Stride time variability	Gait	–	7
5	Acc. standard deviation	Gait	ML	6
6	Stride regularity	Gait	ML	1
7	Stride freq. variability	Gait	AP	1
8	Low freq. percentage	Gait	AP	1
9	Body mass	Other		4
10	Spectral dominant freq.	Gait	AP	1
MULTIPLE-TIME FALL PREDICTION
1	Harmonic ratio	Gait	V	1 (6)
2	Number of falls (past 6 month)	Other	–	5
3	Harmonic ratio	Gait	AP	1
4	Low frequency percentage	Gait	ML	3
5	Spectral freq. range	Gait	ML	3
6	Low frequency percentage	Gait	AP	3
7	Lyapunov W	Gait	AP	1(2)
8	Stride regularity	Gait	AP	1(2)
9	Stride regularity	Gait	ML	1(2)
10	Stride regularity	Gait	V	1(2)

^*Note that values within the brackets indicate that the gait features are equally represented in two latent variables.