Ensembled neural networks applied to modeling survival rate for the patients with out-of-hospital cardiac arrest

The purpose of this study is to use ensembled neural networks (ENN) to model survival rate for the patients with out-of-hospital cardiac arrest (OHCA). We also use seven different sensitivity analyses to find out the important variables to establish a comprehensive and objective assessment method for the OHCA patients. After pre-filtering, we obtained 4,095 data for building this ENN model. The data have been divided into 60?% data for training, 20?% data for validation, and 20?% data for testing. The 11 inputs, including response time, on-scene time, patient transfer time, time to cardiopulmonary resuscitation (CPR), CPR on the scene, using drugs, age, gender, using airway, using automated external defibrillator (AED), and trauma type, and one output variable have been selected as ENN model structure. The results have been shown that ENN can model the OHCA patients and CPR on the scene, using drugs, on-scene time, and using airway in the top 4 of these 11 important variables after 7 different sensitivity analyses. Moreover, these four variables have also been shown significant differences when we use traditional one variable statistics analysis for these variables.