Prediction models in healthcare using deep learning

Abstract

Predictive models are used to predict the unknown future events using a set of relevant predictors or variables by studying both present and historical data. Predictive modeling is also known as predictive analytics that uses the techniques of statistics, data mining, and artificial intelligence that can be applied to a wide set of applications. A predictive model in healthcare learns the historical data of patients to predict their future conditions and determine the treatment. In this review, the use of deep learning models such as LSTM/Bi-LSTM (Long Short-Term Memory/Bi-directional LSTM), RNN (Recurrent Neural Network), CNN (Convolutional Neural Network), RBM (Restricted Boltzmann Machine) and GRU (Gated Recurrent Unit) on different healthcare applications are highlighted. The results indicate that the LSTM/Bi-LSTM model is widely used in time-series medical data and CNN for medical image data. A deep learning model can assist healthcare professionals to make decisions regarding medications, hospitalizations quickly and thus save time and also serve the healthcare industry better. This paper analyzes the various predictive models used in healthcare applications using deep learning.