Real-Time System Prediction for Heart Rate Using Deep Learning and Stream Processing Platforms

Table 4

The result of DL models for lag of 10 minutes and forecasting 15 minutes in advance.

DL models	No. of layers	No. of neurons	Dropout	RMSE
RNN	1	160	0.4	3.113
	2	[100,80]	[0.5,0.1]	3.100
	3	[100,120,40]	[0.2,0.5,0.4]	3.104
LSTM	1	140	0.3	3.043
	2	[70,40]	[0.2,0.2]	3.031
	3	[130,150,110]	[0.4,0.2,0.4]	3.022
GRU	1	110	0.4	3.068
	2	[110,190]	[0.3,0.4]	3.074
	3	[160,60,70]	[0.3,0.4,0.2]	3.043
BI-LSTM	1	40	0.5	3.052
	2	[60,80]	[0.5,0.3]	3.263
	3	[160,60,60]	[0.4,0.4,0.3]	3.0582