Table 4.

Performance across all patients of (i) baseline (Google STT service), (ii) 'fixed' version of NUVA (iii) 'adapted' version of NUVA, (iv) a second independent SLT, SLT2, against the primary SLT scorer used in this study.

System	Accuracy	False Positives	False Negatives	F1-Score	Pearson's r
baseline	0.882	0.020	0.098	0.795	0.727
fixed	0.905	0.049	0.046	0.855	0.784
adapted	0.913	0.057	0.031	0.871	0.807
SLT2	0.965	0.016	0.018	0.947	0.921