基于深度学习的异构时序事件患者数据表示学习框架

刘卢琛; 沈剑豪; 张铭; 王子昌; 李浩然; 刘泽群

doi:10.11959/j.issn.2096-0271.2019003

您当前的位置：

首页 >

文章列表页 >

基于深度学习的异构时序事件患者数据表示学习框架

专题：健康医疗大数据 | 更新时间：2024-06-03

- 基于深度学习的异构时序事件患者数据表示学习框架
- Deep learning based patient representation learning framework of heterogeneous temporal events data
- 大数据 2019年5卷第1期页码：2019003
- 作者机构：
- 作者简介：
  
  [ "刘卢琛（1991- ），男，北京大学信息科学技术学院博士生，主要研究方向为深度学习、医疗大数据等。" ]
  [ "沈剑豪（1995- ），男，北京大学信息科学技术学院博士生，主要研究方向为机器学习、自然语言处理等。" ]
  [ "张铭（1966- ），女，北京大学信息科学技术学院教授、博士生导师，教育部高等学校大学计算机课程教学指导委员会委员，中国计算机学会（CCF）教育工作委员会副主任，ACM教育专家委员会唯一的中国理事，中国ACM教育专家委员会主席。主要研究方向为数据挖掘、机器学习、知识图谱等。" ]
  [ "王子昌（1996- ），男，北京大学信息科学技术学院硕士生，主要研究方向为深度学习、医疗大数据、知识图谱等。" ]
  [ "李浩然（1993- ），男，北京大学信息科学技术学院硕士生，主要研究方向为知识图谱和医疗大数据。" ]
  [ "刘泽群（1997- ），女，北京大学信息科学技术学院本科生，研究兴趣为深度学习、医疗数据挖掘等。" ]
- 基金信息：
  
  国家自然科学基金资助项目;The National Natural Science Foundation of China(No.61772039);国家自然科学基金资助项目;The National Natural Science Foundation of China(No.91646202);国家自然科学基金资助项目;The National Natural Science Foundation of China(No.61472006);北京市科学技术委员会重点研发基金资助项目;Beijing Municipal Commission of Science and Technology Grant(No.Z181100008918005)
- DOI：10.11959/j.issn.2096-0271.2019003
  中图分类号： TP183
- 网络首发：2019-01，
  
  纸质出版：2019-01-15
- 稿件说明：
移动端阅览
刘卢琛, 沈剑豪, 张铭, 等. 基于深度学习的异构时序事件患者数据表示学习框架[J]. 大数据, 2019,5(1):2019003.

Luchen LIU, Jianhao SHEN, Ming ZHANG, et al. Deep learning based patient representation learning framework of heterogeneous temporal events data[J]. Big data research, 2019, 5(1): 2019003.
刘卢琛, 沈剑豪, 张铭, 等. 基于深度学习的异构时序事件患者数据表示学习框架[J]. 大数据, 2019,5(1):2019003. DOI： 10.11959/j.issn.2096-0271.2019003.

Luchen LIU, Jianhao SHEN, Ming ZHANG, et al. Deep learning based patient representation learning framework of heterogeneous temporal events data[J]. Big data research, 2019, 5(1): 2019003. DOI： 10.11959/j.issn.2096-0271.2019003.

摘要

患者数据的表示学习可以将患者历史信息综合表达为一个向量，用于预测未来可能发生的疾病。患者的历史记录可以被建模为多来源数据构成的采样频率差异很大、包含非线性时序关系的异构时序事件。提出了一个新的异构事件长短期记忆表示学习框架，用于学习患者异构时序事件的联合表征。异构事件长短期记忆模型加入了一个可以控制事件访问频率的门，以对不同事件的不规则采样频率建模，同时抓住事件中的复杂时序依赖关系。真实临床数据的实验表明，该方法可以在一系列先进模型的基础上，提升死亡预测和异常实验结果预测的准确度。

Abstract

Patient representation embeds patients' longitude records from multiple sources into continuous low-dimension vectors

which can be used to predict whether a disease will happen in the future. However

the problem is very challenging since patients' history records contain multiple heterogeneous temporal events. The visiting patterns of different types of events vary significantly

and there exist complex nonlinear relationships between different events. A novel model for learning the joint representation of heterogeneous temporal events was proposed. The model adds a new gate to control the visiting rates of different events which effectively models the irregular patterns of different events and their nonlinear correlations. Experiment results with real-world clinical data on the tasks of predicting death and abnormal lab tests prove the effectiveness of the proposed approach over competitive baselines.

关键词

Keywords

references

HE K , ZHANG X , REN S , et al . Deep residuallearning for image recognition [C ] // The IEEE Conference on Computer Vision and Pattern Recognition , June 26 - July 1 , 2016 , Las Vegas, USA .[S.l:s.n ] , 2016 : 770 - 778 .

RAJKOMAR A , OREN Es , CHEN K , et al . Scalable and accurate deep learning with electronic health records [J ] . Digital Medicine , 2018 , 1 ( 1 ): 18 .

BLUNSOM P , CHO K , DYER C , et al . From characters to understanding natural language (c2nlu): robust end-to-end deep learning for nlp (dagstuhl seminar17042) [R ] . 2017 .

GHASSEMI M , PIMENTEL M A , NAUMANN T , et al . A multivariate time series modeling approach to severity of illness assessment and forecasting in icu with sparse, heterogeneous clinical data [C ] // The AAAI Conference on Artificial Intelligence , January 25 - 30 , 2015 , Austin, USA . Palo Alto : AAAI Press , 2015 : 446 - 453 .

NEIL D , PFEIFFER M , LIU S C . Phased LSTM: accelerating recurrent network training for long or event based sequences [C ] // Neural Information Processing Systems 2016 , December 5 - 10 , 2016 , Barcelona, Spain .[S.l:s.n ] , 2016 : 3882 - 3890 .

JOHNSON A E , POLLARD T J , SHEN L , et al . Mimic-iii, a freely accessible critical care database [J ] . Scientific Data , 2016 ( 3 ): 160035 .

BARAJAS K L C , AKELLA R . Dynamicallymodeling patient's health state from electronic medical records: a time series approach [C ] // The 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , August 10 - 13 , 2015 , Sydney, Australia . New York : ACM Press , 2015 : 69 - 78 .

ALAAA M , HU S , SCHAARM V D . Learning from clinical judgments: semi-markovmodulated marked hawkes processes for risk prognosis [J ] . Computer Science , 2017 arXiv:1705. 05267 .

LIU C , WANG F , HU J , et al . Temporal phenotyping from longitudinal electronic health records: a graph based framework [C ] // The 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , August 10 - 13 , 2015 , Sydney, Australia . New York : ACM Press , 2015 : 705 - 714 .

HENRIKSSON A , ZHAO J , BOSTROM H , et al . Modeling electronic health records in ensembles of semantic spaces for adverse drug event detection [C ] // 2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) , November 9 - 12 , 2015 , Washington, DC, USA . Piscataway : IEEE Press , 2015 .

CHOI E , BAHADORI MT , SUN J , et al . Retain: an interpretable predictive modelfor healthcare using reverse time attention mechanism [C ] // Neural Information Processing Systems 2016 , December 5 - 10 , 2016 , Barcelona, Spain .[S.l:s.n ] , 2016 : 3504 - 3512 .

NGUYEN P , TRAN T , WICKRAMASINGHE N , et al . Deepr: a convolutional net for medical records [J ] . IEEE Journal of Biomedical and Health Informatics , 2017 , 21 ( 1 ): 22 - 30 .

HOCHREITER S , BENGIO Y V , FRASCONI P , et al . Gradient flow in recurrent nets: the difficulty of learning long-term dependencies [M ] . Wiley : Wiley-IEEE Press , 2001 .

HOCHREITER S , SCHMIDHUBER J . Long short-termmemory [J ] . Neural Computation , 1997 , 9 ( 8 ): 1735 - 1780 .

KOUTNIK J , GREFF K V , GOMEZ F , et al . A clockwork RNN [C ] // The 31st International Conference on International Conference on Machine Learning , June 21 - 26 , 2014 , Beijing,China .[S.l:s.n ] :JMLR.org, 2014 : 1863 - 1871 .

CAMPOS V , JOU B , GIR-I-NIETO X , et al . Skip RNN: learning to skip state updates in recurrent neural networks [C ] // International Conference on Learning Representations , April 30 - May 3 , 2018 , Vancouver, Canada , 2018 .

BERGSTRA J , BREULEUX O , BASTIEN F , et al . Theano: aCPU and GPU math compiler in Python [C ] // The 9th Python for Scientific Computing Conference , June 28 - July 3 , 2010 , Austin, USA .[S.l:s.n ] , 2010 : 1 - 7 .

KINGMA D , BA J . Adam: a method for stochastic optimization [J ] . Computer Science , 2014 arXiv:1412. 6980 .

TURPIN A , SCHOLERF . User performance versus precision measures for simple search tasks [C ] // The 29th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval , August 6 - 11 , 2006 , Seattle, USA . New York : ACM Press , 2006 : 11 - 18 .

浏览量

966

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于改进YOLOv8的高分辨率遥感图像目标检测算法

沙尘图像视觉增强技术综述

情感语音合成综述

基于生成对抗网络的多特征融合去雾技术

面向非平行语料的语音转换技术综述