HieNet: Bidirectional Hierarchy Framework for Automated ICD Coding

International Classification of Diseases (ICD) is a set classification codes for medical records. Automated ICD coding, which assigns unique with each record, widely used recently its efficiency and error-prone avoidance. However, there are challenges that remain such as heterogeneity, label unbalance, complex relationships between codes. In this work, we proposed novel Bidirectional Hierarchy Framework(HieNet) to address the challenges. Specifically, personalized PageRank routine developed capture co-relation codes, bidirectional hierarchy passage encoder codes’ hierarchical representations, progressive predicting method then narrow down semantic searching space prediction. We validate our on two datasets. Experimental results authoritative public datasets demonstrate boosts state-of-the-art performance by large margin.