Predicting Clinical Events by Combining Static and Dynamic Information Using Recurrent Neural Networks
This work addresses a domain-specific problem for healthcare by improving prediction of post-transplant outcomes, but it is incremental as it adapts existing RNN methods to clinical data.
The authors tackled the problem of predicting clinical events like kidney rejection, loss, or death within 6-12 months after hospital visits by combining static and dynamic patient data from electronic health records, finding that a GRU-based RNN performed best for this task, while a feedforward neural network was better for next event prediction.
In clinical data sets we often find static information (e.g. patient gender, blood type, etc.) combined with sequences of data that are recorded during multiple hospital visits (e.g. medications prescribed, tests performed, etc.). Recurrent Neural Networks (RNNs) have proven to be very successful for modelling sequences of data in many areas of Machine Learning. In this work we present an approach based on RNNs, specifically designed for the clinical domain, that combines static and dynamic information in order to predict future events. We work with a database collected in the Charité Hospital in Berlin that contains complete information concerning patients that underwent a kidney transplantation. After the transplantation three main endpoints can occur: rejection of the kidney, loss of the kidney and death of the patient. Our goal is to predict, based on information recorded in the Electronic Health Record of each patient, whether any of those endpoints will occur within the next six or twelve months after each visit to the clinic. We compared different types of RNNs that we developed for this work, with a model based on a Feedforward Neural Network and a Logistic Regression model. We found that the RNN that we developed based on Gated Recurrent Units provides the best performance for this task. We also used the same models for a second task, i.e., next event prediction, and found that here the model based on a Feedforward Neural Network outperformed the other models. Our hypothesis is that long-term dependencies are not as relevant in this task.