DeepMPM: a mortality risk prediction model using longitudinal EHR data

BMC Bioinformatics

Table 4 The results of the performances of different models

Model	AUC	Precision	Recall	F1-score
RNN	0.8318 ± 0.0102	0.7392 ± 0.0340	0.7571 ± 0.0408	0.7505 ± 0.0139
Multi-task learning	0.6330 ± 0.0084	0.6130 ± 0.0245	0.5808 ± 0.1439	0.5868 ± 0.0674
LSTM-NN	0.8326 ± 0.0087	0.7562 ± 0.0289	0.7508 ± 0.0519	0.7621 ± 0.0148
RETAIN	0.8268 ± 0.0081	0.7592 ± 0.0103	0.7788 ± 0.0091	0.7687 ± 0.0089
Deepcare	0.7876 ± 0.0098	0.7858 ± 0.0264	0.7707 ± 0.0458	0.7782 ± 0.0147
DeepMPM-w/o-\(\beta\)	0.8435 ± 0.0073	0.7685 ± 0.0210	0.7759 ± 0.0490	0.7710 ± 0.0177
DeepMPM	0.8501 ± 0.0076	0.7700 ± 0.0306	0.7987 ± 0.0538	0.7824 ± 0.0153

ISSN: 1471-2105