[Retracted] Automatic Detection of High-Frequency Oscillations Based on an End-to-End Bi-Branch Neural Network and Clinical Cross-Validation

Table 8

Classification performance with different structures.

Group	Model	Confusion matrix					Evaluation metrics (%)
		TN	FP	FN	TP	SEN	SPE	PRE	ACC	FDR	F1
1	Bi-	838	132	50	675	93.10	86.39	83.64	89.26	16.36	88.12
	TF-	866	104	254	471	64.97	89.28	81.91	78.88	18.09	72.46
	Sig-	768	202	69	656	90.48	79.18	76.46	84.01	23.54	82.88
2	Bi-	954	109	120	780	86.67	89.75	87.74	88.33	12.26	87.20
	TF-	980	83	429	471	52.33	92.19	85.02	73.92	14.98	64.79
	Sig-	886	177	92	808	89.78	83.35	82.03	86.30	17.97	85.73
3	Bi-	865	256	30	783	96.31	77.16	75.36	85.21	24.64	84.56
	TF-	974	147	243	570	70.11	86.89	79.50	79.83	20.50	74.51
	Sig-	815	306	22	791	97.29	72.70	72.11	83.04	27.89	82.83
4	Bi-	2282	61	41	2370	98.30	97.40	97.49	97.85	2.51	97.89
	TF-	2308	35	363	2048	84.94	98.51	98.32	91.63	1.68	91.14
	Sig-	2098	245	18	2393	99.25	89.54	90.71	94.47	9.29	94.79
5	Bi-	2642	274	418	2487	85.61	90.60	90.08	88.11	9.92	87.79
	TF-	2656	260	793	2112	72.70	91.08	89.04	81.91	10.96	80.05
	Sig-	2288	628	174	2731	94.01	78.46	81.30	86.22	18.70	87.20