Table 2 The outcomes of prediction models(mean ± std)
From: A time series driven model for early sepsis prediction based on transformer module
Model | Baseline | Time window + RNN | Time window + LSTM | Time window + CNN + Transformer | Time window + LSTM + Transformer |
|---|---|---|---|---|---|
Accuracy | 0.637 (0.035) | 0.653 (0.032) | 0.835 (0.023) | 0.951 (0.017) | 0.964 (0.018) |
Precision | 0.711 (0.037) | 0.715 (0.030) | 0.875 (0.047) | 0.955 (0.028) | 0.956 (0.012) |
Recall | 0.677 (0.044) | 0.691 (0.041) | 0.861 (0.011) | 0.951 (0.012) | 0.967 (0.012) |
F1-score | 0.667 (0.045) | 0.662 (0.039) | 0.871 (0.054) | 0.954 (0.023) | 0.959 (0.014) |