Hyperbolic graph convolutional neural network with contrastive learning for automated ICD coding

Yuzhou Wu, Xuechen Chen, Xin Yao, Yongang Yu, Zhigang Chen

Abstract: Highlights•The aim is to classify EHRs with the International Classification of Diseases.•A contrastive learning framework is proposed for automated ICD coding.•Adopting Hyperbolic Graph Convolutional Networks can drop distortion of hierarchichy.•The model achieved competitive results compared to state-of-the-art models.