Table 2.

Summarized spindle detection performance of SpindleNet on the MASS dataset (n=19 subjects).

MASS subject #	Cross validation fold #	Median latency (ms)	Sensitivity	Specificity	FDR	F1-score	AUROC	False event rate (spindles/min)	True event rate (spindles/min)
1	1	250.85	0.8739	0.9888	0.0566	0.9072	0.9944	0.5517	9.1818
2		250.16	0.927	0.9829	0.1088	0.9087	0.9951	0.8892	7.2782
3		286.44	0.7142	0.9971	0.0572	0.8127	0.9954	0.1633	2.6917
4		215.09	0.9353	0.9633	0.2682	0.8211	0.99	3.5114	1.0182
5	2	269.76	0.7153	0.9919	0.1418	0.7802	0.9902	1.9859	5.4174
6		262.96	0.89	0.9672	0.1904	0.8479	0.9864	0.4571	2.765
7		355.86	0.7882	0.9926	0.0601	0.8574	0.9906	1.7033	7.2412
8		342.94	0.7355	0.9987	0.0096	0.8441	0.9953	6.3961	2.11
9	3	238.9	0.954	0.9597	0.2215	0.8574	0.9923	0.3857	6.0288
10		273.01	0.8854	0.9805	0.1681	0.8579	0.9912	0.0647	6.6039
11		227.11	0.9711	0.9414	0.3486	0.7798	0.9909	2.1067	7.4065
12		179.2	0.9774	0.9018	0.4145	0.7323	0.9807	1.0538	5.2166
13	4	333.87	0.9087	0.9829	0.1761	0.8642	0.9944	3.1616	5.9087
14		207.54	0.9444	0.9491	0.2721	0.8222	0.9888	5.1579	7.2868
15		223.6	0.9604	0.9609	0.3601	0.768	0.9927	2.4406	0.5715
16		47.91	0.9605	0.9404	0.7752	0.3643	0.976	4.0902	2.5528
17	5	58.425	0.9974	0.8895	0.7519	0.4977	0.9808	0.9449	4.4194
18		118.88	0.9897	0.9589	0.8102	0.3184	0.9938	2.6696	7.1434
19		114.9	0.9845	0.9288	0.6157	0.5528	0.9858	2.1905	3.8921