Research of GPCEMBP Glucose Prediction Algorithm Based on Continuous Glucose Monitoring

Table 1

Selection of hidden layer and neurons in the BP neural network.

pH (min)	Inputs	Network architecture	Output	FIT (%)	RMSE (mg/dl)	NPE (%)	ET (s)
63	12	20-1	1	90.38	16.87	8.1	7.85
63	12	12-8-1	1	93.42	14.07	6.75	10.14
63	12	10-6-4-1	1	92.95	14.55	6.99	18.57
63	12	28-1	1	91.85	15.59	7.48	12.09
63	12	15-13-1	1	95.66	11.51	5.52	16.78
63	12	10-10-8-1	1	94.85	12.49	6.08	15.36
63	12	35-1	1	92.31	15.17	7.28	21.25
63	12	20-15-1	1	97.4	8.94	4.29	27.64
63	12	15-12-8-1	1	97.63	8.61	4.13	24.92
63	12	45-1	1	91.16	14.77	7.09	21.41
63	12	25-20-1	1	98.32	7.22	3.47	29.10
63	12	20-13-12-1	1	96.10	10.91	5.24	39.70