Research Article

[Retracted] Deep Learning-Based Emergency Care Process Reengineering of Interventional Data for Patients with Emergency Time-Series Events of Myocardial Infarction

Table 3

Experimental results for prediction of death and prediction of abnormal laboratory results.

MethodDeath predictionPotassium ion anomaly prediction
AUCAPAUCAP

Independent LSTM0.87710.55730.71960.2969
Independent LSTM with shared parameters0.80640.53010.53080.1098
Phased LSTM0.84740.49000.77220.3575
Clock-work RNN0.84000.71810.65160.2208
Retain0.89670.58080.73250.3096
LSTM + event embedding0.94660.74450.72310.3021
HE-LSTM0.95160.76870.79870.3914