Graph-Augmented Cyclic Learning Framework for Similarity Estimation of Medical Clinical Notes

TL;DR

This paper introduces a graph-augmented cyclic learning framework that enhances clinical note similarity estimation by integrating domain knowledge with language models, significantly improving performance over baseline models.

Contribution

The novel framework combines language models with a GCN-based network for domain knowledge infusion, improving clinical text similarity estimation.

Findings

Improved Bio-clinical BERT baseline by 16.3%

Enhanced performance with co-training by 27.9%

Effective integration of domain knowledge in clinical NLP

Abstract

Semantic textual similarity (STS) in the clinical domain helps improve diagnostic efficiency and produce concise texts for downstream data mining tasks. However, given the high degree of domain knowledge involved in clinic text, it remains challenging for general language models to infer implicit medical relationships behind clinical sentences and output similarities correctly. In this paper, we present a graph-augmented cyclic learning framework for similarity estimation in the clinical domain. The framework can be conveniently implemented on a state-of-art backbone language model, and improve its performance by leveraging domain knowledge through co-training with an auxiliary graph convolution network (GCN) based network. We report the success of introducing domain knowledge in GCN and the co-training framework by improving the Bio-clinical BERT baseline by 16.3% and 27.9%, respectively.