Prediction of Hospital Readmission for Heart Disease: A Deep Learning Approach

  • Jingwei Da
  • , Danni Yan
  • , Sijia Zhou
  • , Yidi Liu
  • , Xin Li
  • , Yani Shi
  • , Jiaqi Yan
  • , Zhongmin Wang

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

Abstract

Hospital readmissions consume large amounts of medical resources and negatively impact the healthcare system. Predicting the readmission rate early one can alleviate the financial and medical consequences. Most related studies only select the patient’s structural features or text features for modeling analysis, which offer an incomplete picture of the patient. Based on structured data (including demographic data, clinical data, administrative data) and medical record text, this paper uses deep learning methods to construct an optimal model for hospital readmission prediction, tested on a dataset of heart disease patients’ 30-day readmission. The results show that when only structured data is used, the deep learning model is much better than the Naive Bayes model and slightly better than the Support Vector Machine model. Adding a text model to the deep learning model improves performance, increasing accuracy and F1-score by 2% and 6%, respectively. This indicates that textual information contributes greatly to hospital readmission predictions.

Original languageEnglish
Title of host publicationSmart Health - International Conference, ICSH 2019, Proceedings
EditorsHsinchun Chen, Daniel Zeng, Xiangbin Yan, Chunxiao Xing
PublisherSpringer
Pages16-26
Number of pages11
ISBN (Print)9783030344818
DOIs
Publication statusPublished - 2019
Externally publishedYes
Event7th International Conference for Smart Health, ICSH 2019 - Shenzhen, China
Duration: 1 Jul 20192 Jul 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11924 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference7th International Conference for Smart Health, ICSH 2019
Country/TerritoryChina
CityShenzhen
Period1/07/192/07/19

Keywords

  • Deep learning
  • Hospital readmission
  • Predictive analytics

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Prediction of Hospital Readmission for Heart Disease: A Deep Learning Approach'. Together they form a unique fingerprint.

Cite this