Table 4.

Results achieved in the CHB-MIT dataset using LSTM and the multi-frame network.

Subject	No.Seia	CNN			multi-frame
		(instance-based)			network
		Acc	Sen	Spe	Acc	Sen	Spe
		(%)	(%)	(%)	(%)	(%)	(%)
chb01	7	89.90	86.82	92.99	96.24	95.34	97.14
chb02	3	84.22	73.09	95.34	88.13	78.26	97.99
chb03	6	74.70	61.04	88.36	85.81	79.29	92.33
chb04	3	76.00	81.39	70.61	78.73	89.13	68.32
chb05	5	60.80	55.90	65.69	77.77	72.53	83.00
chb07	3	87.98	83.43	92.53	94.34	91.00	97.68
chb08	5	69.29	78.45	57.87	77.89	82.04	72.70
chb09	4	59.94	54.47	65.41	65.84	51.01	80.67
chb10	6	55.99	47.80	64.18	64.54	61.90	67.19
chb13	5	85.58	81.94	89.22	95.24	94.91	95.57
chb14	5	53.24	46.18	60.31	56.89	60.59	53.18
chb16	6	50.19	41.64	58.74	57.36	54.89	59.84
chb17	3	89.21	86.04	92.39	93.35	89.00	97.70
chb18	6	83.19	84.34	82.05	89.90	89.56	90.25
chb19	3	73.10	57.52	88.69	77.44	61.98	92.89
chb20	5	78.78	68.51	89.04	97.82	97.47	98.17
chb21	4	57.83	43.45	72.21	67.43	58.13	76.74
chb23	5	87.68	92.50	83.20	92.63	97.69	87.94
Mean		73.20	68.03	78.27	80.96	78.04	83.85

^aNumber of seizures.