Deep Representation for Patient Visits from Electronic Health Records

TL;DR

This paper introduces a deep learning approach to generate low-dimensional embeddings of patient visits from electronic health records, excluding diagnosis codes, to enhance predictive modeling in personalized medicine.

Contribution

It presents a novel method for learning meaningful patient visit embeddings using deep neural networks trained without ICD codes, facilitating improved healthcare predictions.

Findings

Embeddings capture relevant clinical information.

Embeddings improve ICD code prediction accuracy.

Medical information aligns with specific embedding directions.

Abstract

We show how to learn low-dimensional representations (embeddings) of patient visits from the corresponding electronic health record (EHR) where International Classification of Diseases (ICD) diagnosis codes are removed. We expect that these embeddings will be useful for the construction of predictive statistical models anticipated to drive personalized medicine and improve healthcare quality. These embeddings are learned using a deep neural network trained to predict ICD diagnosis categories. We show that our embeddings capture relevant clinical informations and can be used directly as input to standard machine learning algorithms like multi-output classifiers for ICD code prediction. We also show that important medical informations correspond to particular directions in our embedding space.