Table 3.

Showing the differences in calculation of trigger delay derived from the formula, operator decision and the deep learning

Trigger delay (ms)	Formula vs operator			Formula vs deep learning			Operator vs deep learning
	Mean diff	P-value	95% CI of diff	Mean diff	P-value	95% CI of diff	Mean diff	P-value	95% CI of diff
All	73 ± 99	< 0.01	36, 110	46 ± 110	0.02	9, 84	− 26 ± 61	0.02	− 49, − 4
Healthy	37 ± 93	0.04	2, 72	35 ± 131	0.16	− 14, 84	− 2 ± 56	0.81	− 23, 18
Patient (standardised AW)	62 ± 72	< 0.01	36, 89	52 ± 75	< 0.01	23, 80	− 11 ± 27	0.04	− 21, − 1
Patient (variable AW)	118 ± 110	< 0.01	77, 159	53 ± 87	< 0.01	20, 85	− 66 ± 69	< 0.01	− 91, − 40

	Formula			Operator			Deep learning
	Mean diff	P-value	95% CI of diff	Mean diff	P-value	95% CI of diff	Mean diff	P-value	95% CI of diff
Healthy vs patient (all)	− 112 (35)	< 0.01	− 184, − 40	− 59 (50)	0.26	− 166, 48	− 95 (60)	0.14	− 224, 34
Patient (standardised AW) vs patient (variable AW)	− 65 (55)	0.26	− 183, 54	− 9 (47)	0.86	− 108, 91	− 63 (53)	0.25	− 176, 49

This is further compared between the whole cohort and the healthy and patient participants. The mean difference was analysed using a one-sample t-test and presented as mean ± SD. Unpaired data was analysed using Welch’s t test and presented as mean and standard error. AW = Acquisition Window, CI of Diff = Confidence Interval of Difference. Statistically significant values (p<0.05) are indicated in bold