Research Article
Intelligent Ensemble Deep Learning System for Blood Glucose Prediction Using Genetic Algorithms
Table 4
Mean (standard deviation in parentheses) RMSE for 29 diabetic patients using univariate and multivariate time-series models.
| | RMSE | PH = 15 min | PH = 30 min | PH = 60 min | | Uni | Multi | Uni | Multi | Uni | Multi |
| | RNN | 11.59 (3.59) | 11.39 (3.43) | 20.2 (6) | 19.45 (5.4) | 32.65 (9.44) | 31.25 (8.94) | | LSTM | 11.89 (3.83) | 11.83 (3.3) | 21.3 (6.05) | 20.59 (6.15) | 33.43 (9.59) | 32.71 (9.7) | | Stacked LSTM | 12.57 (5.39) | 11.59 (3.43) | 20.57 (5.66) | 20.54 (6.12) | 34.63 (9.98) | 31.91 (8.9) | | Bidirectional LSTM | 12.03 (3.42) | 12.02 (3.6) | 21.01 (6.05) | 19.89 (5.63) | 33.54 (9.25) | 31.99 (9.24) | | GRU | 11.50 (3.54) | 11.37 (3.29) | 20.46 (5.89) | 19.86 (5.4) | 32.96 (9.36) | 32.25 (9.12) | | Baseline ARIMA | 14.82 (4.41) | | 23.11 (6.66) | | 35.67 (10.23) | | | Proposed model | 11.28 (3.34) | 11.08 (3.19) | 19.99 (5.59) | 19.25 (5.28) | 33.13 (9.27) | 31.3 (8.81) | | -value | | | | | 0.32 | |
|
|
|
