Table 4.

Confusion matrix and overall performance on the sleep-edfx dataset.

Stage	Predictions					Per-Class Metrics
	W	N1	N2	N3	REM	Precision	Recall	F1
W	7287	586	89	57	149	91.5	89.2	90.3
N1	279	1497	434	24	570	42.1	53.4	47.1
N2	259	846	14,596	1388	710	90.5	82.1	86.0
N3	39	31	586	5042	5	76.6	88.4	82.1
REM	103	598	422	69	6525	82.0	84.6	83.2

Overall accuracy: 82.8%, MF1 score: 77.8.