$\mathtt{MedGraph:}$ Structural and Temporal Representation Learning of Electronic Medical Records

TL;DR

MedGraph is a novel supervised embedding method for electronic medical records that captures rich visit and code attributes, as well as temporal dynamics, leading to improved performance in medical risk prediction.

Contribution

It introduces MedGraph, which models visit-code relationships and temporal visit sequences using attributed bipartite graphs and point processes, with Gaussian embeddings to handle uncertainty.

Findings

Outperforms existing EMR embedding methods in risk prediction tasks.

Effectively captures attribute information and temporal dynamics.

Provides uncertainty modeling through Gaussian embeddings.

Abstract

Electronic medical record (EMR) data contains historical sequences of visits of patients, and each visit contains rich information, such as patient demographics, hospital utilisation and medical codes, including diagnosis, procedure and medication codes. Most existing EMR embedding methods capture visit-code associations by constructing input visit representations as binary vectors with a static vocabulary of medical codes. With this limited representation, they fail in encapsulating rich attribute information of visits (demographics and utilisation information) and/or codes (e.g., medical code descriptions). Furthermore, current work considers visits of the same patient as discrete-time events and ignores time gaps between them. However, the time gaps between visits depict dynamics of the patient's medical history inducing varying influences on future visits. To address these limitations, we present $\mathtt{MedGraph}$, a supervised EMR embedding method that captures two types of information: (1) the visit-code associations in an attributed bipartite graph, and (2) the temporal sequencing of visits through a point process. $\mathtt{MedGraph}$ produces Gaussian embeddings for visits and codes to model the uncertainty. We evaluate the performance of $\mathtt{MedGraph}$ through an extensive experimental study and show that $\mathtt{MedGraph}$ outperforms state-of-the-art EMR embedding methods in several medical risk prediction tasks.