Table 3.

Measures of FOG occurrences quantified in the two test sets and estimated by the winning models (mean ± SD)

Private test: n = 14 Private+Public test: n = 40		% Time frozen	No. of FOG episodes	FOG duration (s)
Gold standard, video-based	Private test	14.39 ± 15.07	27.93 ± 22.23	184.12 ± 353.49
	Private+public test	22.64 ± 20.98	33.25 ± 25.07	273.66 ± 346.89
1st place	Private test	12.44 ± 16.88	16.36 ± 18.41	175.54 ± 374.25
	Private+public test	23.11 ± 23.38	21.00 ± 21.45	268.43 ± 351.53
2nd place	Private test	11.65 ± 17.98	27.71 ± 31.31	173.08 ± 371.49
	Private+public test	19.33 ± 22.16	39.87 ± 41.65	235.28 ± 337.64
3rd place	Private test	13.65 ± 18.05	37.21 ± 45.99	194.94 ± 401.92
	Private+public test	24.37 ± 24.72	47.80 ± 38.68	296.80 ± 391.58
4th place	Private test	13.27 ± 17.99	130.71 ± 136.82	175.42 ± 393.58
	Private+public test	24.21 ± 23.62	164.70 ± 175.38	282.60 ± 377.09
5th place	Private test	15.67 ± 16.24	28.64 ± 26.13	185.41 ± 367.78
	Private+public test	20.28 ± 20.86	37.97 ± 37.64	224.81 ± 320.78

Median values and interquartile ranges (IQRs) can be found in Supplementary Table 3.