Predicting Hospital Length of Stay using Neural Networks on MIMIC III Data

In this paper we explore the use of neural networks for predicting the total length of stay for patients with various diagnoses based on selected general characteristics. A neural network is trained to predict whether patient stay will be long (> 5 days), or short (<= 5 days) as of the time the patient leaves the ICU unit. Our dataset is drawn from the MIMIC III database and all code was written in R and in Postgress, while the computations were executed on the Florida Polytechnic University's supercomputer. Our prediction accuracy is approximately 80% and clearly outperforms any linear model.