A Novel Dynamic Update Framework for Epileptic Seizure Prediction

Table 1

Comparison of no update and dynamic update for the model.

Patient	No update for the model					Dynamic update for the model
	Sensitivity (%)	Advance times (min)			False-positive  rate (h^−1)	Sensitivity (%)	Advance times (min)			False-positive rate (h^−1)
		1	2	3			1	2	3
4	100	79.1	58.7	62.6	0.03	100	79.0	56.2	63.5	0.00
5	66.7	0	24.3	43.8	0.00	66.7	17.5	0	22.0	0.03
9	100	67.8	59.2	81.3	0.00	100	67.8	59.2	81.3	0.00
10	100	18.4	27.3	76.0	0.00	100	20.3	27.1	76.4	0.07
16	66.7	73.3	10.9	0	0.03	66.7	45.0	23.8	0	0.00
17	33.3	108.3	0	0	0.38	66.7	52.8	0	33.3	0.14
18	100	38.3	63.3	93.8	0.11	100	38.3	68.5	93.7	0.00
20	66.7	0	26.9	36.9	0.17	66.7	0	30.9	37.5	0.12
21	100	108.2	102.5	76.0	0.21	100	107.2	103.0	61.5	0.00