Using Markov Models to Find Interesting Patient Pathways

Sally McClean, L Garg, Brian Meenan, Peter Millard

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

Over recent years the concept of Interestingness has come to underpin Data Mining, leading to the discovery of much new knowledge. In particular recognition of interesting patient pathways can lead to the discovery of important rules and patterns such as high probability pathways, groups of patients who incur exceptional high costs or pathways that are very long lasting. In the current paper we show how Markov models can be used to identify such patient pathways. Using Markov modelling we show how patient pathways may be extracted and describe an algorithm based on branch and bound that we have developed to efficiently extract a number of interesting pathways, subject to the number of pathways required, or some other criterion being specified. The approach is illustrated using data on geriatric patients from an administrative database of a London hospital, and we identify interesting pathways for geriatric patients. Such an approach might be used in association with healthcare process improvement technologies, such as Lean Thinking or Six Sigma.
Original languageEnglish
Title of host publicationUnknown Host Publication
PublisherIEEE
Pages713-718
Number of pages6
ISBN (Print)0-7695-2905-4
DOIs
Publication statusPublished - 2007
EventTwentieth IEEE International Symposium on Computer-Based Medical Systems (CBMS'07) - Maribor, Slovenia
Duration: 1 Jan 2007 → …

Conference

ConferenceTwentieth IEEE International Symposium on Computer-Based Medical Systems (CBMS'07)
Period1/01/07 → …

Fingerprint Dive into the research topics of 'Using Markov Models to Find Interesting Patient Pathways'. Together they form a unique fingerprint.

  • Cite this

    McClean, S., Garg, L., Meenan, B., & Millard, P. (2007). Using Markov Models to Find Interesting Patient Pathways. In Unknown Host Publication (pp. 713-718). IEEE. https://doi.org/10.1109/CBMS.2007.121