Table 2 Comparative results of different representation learning methods for mortality prediction
From: Representation learning for clinical time series prediction tasks in electronic health records
AUC | Threshold =0.8 | ||
|---|---|---|---|
Accuracy | F1-score | ||
RNN-DAE | 0.783 | 0.779 | 0.449 |
RNN-AE | 0.755 | 0.760 | 0.444 |
SDAs | 0.755 | 0.738 | 0.439 |
Hand | 0.608 | 0.693 | 0.420 |
PCA | 0.602 | 0.715 | 0.427 |
GMM | 0.607 | 0.693 | 0.420 |
k-means | 0.628 | 0.722 | 0.430 |