Table 1.

Features extracted from windowed data.

Feature	Feature Description	Source	Number of Input Parameters	Total Features
Time domain features (n = 13)
Number, duration, length of COP reversals	Number, length, duration of centre of pressure (COP) path direction reversals per window (n = 3)	[36]	2	6
Number, duration, length of COP deviations	Number, length, duration of mediolateral COP deviations per window. Deviation is the first derivative of COP ML exceeding a threshold of ±0.5 mm/window (n = 3)	[36]	2	6
CV of COP position, velocity, acceleration	Anterior/posterior (AP) and medial/lateral (ML) coefficients of variation (CV) of COP position, velocity, and acceleration (n = 6)	[36]	2	12
Number of weight shifts	Number of times the majority of total GRF (>50%) changed foot (n = 1)	-	1	1
			Category total	25
Fast Fourier transform (FFT) features (n = 8)
Total power in FFT signal	Power in FFT signal per window as sum of squared amplitude (n = 1)	[37]	38	38
Dominant frequency	Frequency bin with highest amplitude per window (n = 1)	[38]	38	38
Max, min, mean	Maximum, minimum, and mean amplitude of FFT signal (n = 3)	[38]	38	114
Power in locomotion, freeze bands	Power under FFT curve in locomotion band (0.5–3 Hz) and freeze band (3–8 Hz) (n = 2)	[39]	38	76
Freeze index	Ratio of power in freeze band (3–8 Hz) and locomotion band (0.5–3 Hz) (n = 1)	[39]	38	38
			Category total	304
Discrete wavelet transform features (n = 14), Haar mother wavelet
Variance of coefficients	Variance of the detail and approximation coefficient vectors (n = 2)	[40]	38	76
Max, min, mean	Maximum, minimum, mean of detail and approximation coefficient vectors (n = 6)	[40]	38	228
Max, min, mean energy	Maximum, minimum, mean energy of detail and approximation coefficient vectors (n = 6)	[40]	38	228
			Category total	532