Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Prediction of re-admissions for critical health conditions: A Machine Learning Approach
Blekinge Institute of Technology, Faculty of Computing, Department of Computer Science and Engineering.
2015 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Context. Re-admission is the return hospitalization within 30 days from the date of original admission or discharge from hospital. Thecosts of the unplanned re-admissions were estimated to $25 billion per year alone in the U.S. Re-admission rate also has a huge impact onquality of care provided to the patients, cost of health care, and utilization of hospital resources and the image of the care provider. Studies indicate huge potential of savings that can be achieved with incremental performance improvements in detecting cases of preventable re-admissions.

Objectives. In this study we find the different features that helpin predicting readmissions, compare different machine learning techniques and build a model to predict readmissions using one technique.We also propose a framework for implementation of this model in the real world situations.

Methods. To reach the objective, the data of the patients over a period of time were studied to determine the factors that help in identifying re-admissions. Experiments are performed for identifying the features that are more relevant to predict re-admissions and for investigating the most suitable machine learning techniques for this purpose.This model was tested to predict re-admission cases for Acute Myocardial Infarction and Pneumonia.

Results. The features that help in predicting re-admissions are determined,and a model was developed using these features and the selected machine learning algorithm. The model showed good results in predicting re-admissions. The model predicted risk of Acute Myocardial Infarction(c-statistic=0.811), and Pneumonia(c=0.76).

Conclusions. We conclude that our model showed good results in predicting re-admissions. The developed model is discriminative for specific diseases like Acute Myocardial Infarction and Pneumonia. Itis also generalized as it incorporates the features that can be easily available from all of the patient population over the globe.

Place, publisher, year, edition, pages
2015.
Keyword [en]
machine learning, prediction, re-admission, readmission
National Category
Computer Science
Identifiers
URN: urn:nbn:se:bth-10821OAI: oai:DiVA.org:bth-10821DiVA: diva2:861506
Subject / course
DV2566 Master's Thesis (120 credits) in Computer Science
Educational program
DVAXA Master of Science Programme in Computer Science
Supervisors
Examiners
Available from: 2015-10-19 Created: 2015-10-17 Last updated: 2015-10-19Bibliographically approved

Open Access in DiVA

fulltext(549 kB)323 downloads
File information
File name FULLTEXT02.pdfFile size 549 kBChecksum SHA-512
090ef1f032cdc158e54fe2507d7df9a73eeaeaa4454722fac778f2114f3bebc4ca3cf33dc3ecf58ae0a7b4c40f8e2d4c4933ab7f71c016e2e1eee7ccb77385e7
Type fulltextMimetype application/pdf

By organisation
Department of Computer Science and Engineering
Computer Science

Search outside of DiVA

GoogleGoogle Scholar
Total: 323 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 690 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf