TACCO: Task-guided Co-clustering of Clinical Concepts and Patient Visits for Disease Subtyping based on EHR Data

TL;DR

TACCO is a novel self-supervised hypergraph co-clustering framework that jointly discovers disease subtypes by clustering clinical concepts and patient visits, improving disease risk prediction from EHR data.

Contribution

It introduces a hypergraph-based co-clustering method guided by disease prediction tasks, integrating textual embeddings and contrastive learning for better disease subtyping.

Findings

Achieved 31.25% performance improvement over traditional ML methods.

Improved cardiovascular risk prediction by 5.26% over baseline hypergraph models.

Validated utility through clinical case studies and detailed analysis.

Abstract

The growing availability of well-organized Electronic Health Records (EHR) data has enabled the development of various machine learning models towards disease risk prediction. However, existing risk prediction methods overlook the heterogeneity of complex diseases, failing to model the potential disease subtypes regarding their corresponding patient visits and clinical concept subgroups. In this work, we introduce TACCO, a novel framework that jointly discovers clusters of clinical concepts and patient visits based on a hypergraph modeling of EHR data. Specifically, we develop a novel self-supervised co-clustering framework that can be guided by the risk prediction task of specific diseases. Furthermore, we enhance the hypergraph model of EHR data with textual embeddings and enforce the alignment between the clusters of clinical concepts and patient visits through a contrastive objective. Comprehensive experiments conducted on the public MIMIC-III dataset and Emory internal CRADLE dataset over the downstream clinical tasks of phenotype classification and cardiovascular risk prediction demonstrate an average 31.25% performance improvement compared to traditional ML baselines and a 5.26% improvement on top of the vanilla hypergraph model without our co-clustering mechanism. In-depth model analysis, clustering results analysis, and clinical case studies further validate the improved utilities and insightful interpretations delivered by TACCO. Code is available at https://github.com/PericlesHat/TACCO.