arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.02797cs.CLcs.IR
收藏

面向自动医疗编码的深度学习统一综述

A Unified Review of Deep Learning for Automated Medical Coding

Aalto University(阿尔托大学);University of Helsinki(赫尔辛基大学);Dalian Maritime University(大连海事大学);KU Leuven(鲁汶大学);University of Exeter(埃克塞特大学);Institute for Molecular Medicine Finland (FIMM)(芬兰分子医学研究所);Terveystalo Healthcare Services(Terveystalo医疗服务公司);School of Health and Wellbeing, University of Glasgow(格拉斯哥大学健康与福祉学院)

Shaoxiong Ji, Wei Sun, Xiaobo Li, Hang Dong, Ara Taalas, Yijia Zhang, Honghan Wu, Esa Pitkänen, Pekka Marttinen

AI总结:

该综述针对基于深度学习的自动医疗编码缺乏统一架构视角的问题,提出包含四大组件的统一框架以梳理现有模型,并介绍基准、应用及未来研究方向。

摘要:

自动医疗编码是医疗运营与服务中的核心任务,它通过从临床文档中预测医疗编码,让非结构化数据变得可管理。近年来深度学习与自然语言处理领域的进展已被广泛应用于该任务,但基于深度学习的医疗编码研究缺乏对神经网络架构设计的统一视角。本综述提出一个统一框架,以帮助学界通用地理解医疗编码模型的构建模块,并在该框架下梳理了近年的先进模型。我们的统一框架将医疗编码拆解为四大核心组件:用于文本特征提取的编码器模块、构建深度编码器架构的机制、将隐藏表示转换为医疗编码的解码器模块,以及辅助信息的使用方式。最后,本文介绍了相关基准数据集与实际应用场景,并讨论了关键研究挑战与未来方向。

英文摘要:

Automated medical coding, an essential task for healthcare operation and delivery, makes unstructured data manageable by predicting medical codes from clinical documents. Recent advances in deep learning and natural language processing have been widely applied to this task. However, deep learning-based medical coding lacks a unified view of the design of neural network architectures. This review proposes a unified framework to provide a general understanding of the building blocks of medical coding models and summarizes recent advanced models under the proposed framework. Our unified framework decomposes medical coding into four main components, i.e., encoder modules for text feature extraction, mechanisms for building deep encoder architectures, decoder modules for transforming hidden representations into medical codes, and the usage of auxiliary information. Finally, we introduce the benchmarks and real-world usage and discuss key research challenges and future directions.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑