Learning Patient Representations from Text

TL;DR

This paper introduces a neural network model for learning comprehensive patient representations from electronic health records, achieving state-of-the-art results in phenotyping tasks like comorbidity detection.

Contribution

The paper presents a novel neural network approach for patient representation learning that outperforms traditional sparse methods in phenotyping tasks.

Findings

Learned representations achieve state-of-the-art performance

Neural network models capture richer patient information

Improved phenotyping accuracy over bag-of-words methods

Abstract

Mining electronic health records for patients who satisfy a set of predefined criteria is known in medical informatics as phenotyping. Phenotyping has numerous applications such as outcome prediction, clinical trial recruitment, and retrospective studies. Supervised machine learning for phenotyping typically relies on sparse patient representations such as bag-of-words. We consider an alternative that involves learning patient representations. We develop a neural network model for learning patient representations and show that the learned representations are general enough to obtain state-of-the-art performance on a standard comorbidity detection task.