Deep Label-Wise Attentive Temporal Convolutional Networks Improve Medical Coding

2026-07-27

AI Summary

This study addresses the challenge of automatically assigning multiple diagnosis and procedure codes from lengthy clinical notes, which requires modeling long-range dependencies while focusing on label-relevant local contexts. Framing medical coding as a multi-label text classification task, the authors propose a novel architecture that integrates deep temporal convolutional networks (TCNs) with a label-wise attention mechanism. The TCN component effectively captures global sequential dependencies across the entire document, while the label-wise attention dynamically directs the model to attend to text segments most relevant to each specific code. Evaluated on standard benchmark datasets, the proposed method substantially outperforms current state-of-the-art models, achieving a 9% absolute improvement in F1 score and a 28% gain in recall, thereby significantly enhancing the practical utility of clinical decision support systems.

Abstract

Medical coding is the task of assigning a set of diagnosis and procedure codes for a hospitalization using recorded notes. It requires aggregating information from different parts of the text and focus to different sections for each individual code, making it a very difficult problem even for professional human coders. We model the task as a multi-label text classification problem. To overcome the mentioned difficulties, we propose a deep neural model consisting of a multi-layer temporal convolution network (TCN) followed by label-wise attention. While multi-layer TCN helps extract a global document representation with the ability to learn relations over very long sequences, label-specific attention mechanism allows the model to focus on different aspects of the same document for each individual label. Our method achieves significantly better F-1 scores (9% increase) compared to the previous state-of-the-art model, with a remarkable increase in recall score (28% increase), which we believe is the more important metric for a clinical decision support setting.

Problem

• medical coding
• multi-label text classification
• diagnosis codes
• procedure codes
• clinical decision support

Innovation

• Temporal Convolutional Network
• Label-wise Attention
• Medical Coding
• Multi-label Text Classification
• Deep Neural Model