Table 3.

Confusion matrix of the combined retrain transfer learning model for sleep stage classification.

Predicted label → ↓ True label	Wake	N1/N2	N3	REM	Sensitivity (%)
Wake	8,468	2,588	199	885	69.75
N1/N2	2,670	31,449	4,332	3,347	75.24
N3	68	6,127	14,570	182	69.56
REM	461	2,298	103	14,690	83.69
PPV (%)	72.58	74.06	75.87	76.89

Note: Results were obtained after classifying Wake, REM sleep, N1/N2 sleep, and N3 sleep in all epochs (n = 92,437) of all recordings on the Eindhoven PPG data set. Sensitivity and positive predictive value (PPV) for each class are presented.