Change search
ReferencesLink to record
Permanent link

Direct link
Temporal pattern discovery in longitudinal electronic patient records
Stockholm University, Faculty of Science, Department of Mathematics.
Show others and affiliations
2010 (English)In: Data mining and knowledge discovery, ISSN 1384-5810, E-ISSN 1573-756X, Vol. 20, no 3, 361-387 p.Article in journal (Refereed) Published
Abstract [en]

Large collections of electronic patient records provide a vast but still underutilised source of information on the real world use of medicines. They are maintained primarily for the purpose of patient administration, but contain a broad range of clinical information highly relevant for data analysis. While they are a standard resource for epidemiological confirmatory studies, their use in the context of exploratory data analysis is still limited. In this paper, we present a framework for open-ended pattern discovery in large patient records repositories. At the core is a graphical statistical approach to summarising and visualising the temporal association between the prescription of a drug and the occurrence of a medical event. The graphical overview contrasts the observed and expected number of occurrences of the medical event in different time periods both before and after the prescription of interest. In order to effectively screen for important temporal relationships, we introduce a new measure of temporal association, which contrasts the observed-to-expected ratio in a time period immediately after the prescription to the observed-to-expected ratio in a control period 2 years earlier. An important feature of both the observed-to-expected graph and the measure of temporal association is a statistical shrinkage towards the null hypothesis of no association, which provides protection against highlighting spurious associations. We demonstrate the usefulness of the proposed pattern discovery methodology by a set of examples from a collection of over two million patient records in the United Kingdom. The identified patterns include temporal relationships between drug prescriptions and medical events suggestive of persistent and transient risks of adverse events, possible beneficial effects of drugs, periodic co-occurrence, and systematic tendencies of patients to switch from one medication to another.

Place, publisher, year, edition, pages
2010. Vol. 20, no 3, 361-387 p.
Keyword [en]
Temporal pattern discovery, Longitudinal patient records, Electronic health records
National Category
Research subject
URN: urn:nbn:se:su:diva-50142DOI: 10.1007/s10618-009-0152-3ISI: 000276276100003OAI: diva2:380560
authorCount :5Available from: 2010-12-21 Created: 2010-12-21 Last updated: 2010-12-21Bibliographically approved

Open Access in DiVA

No full text

Other links

Publisher's full text

Search in DiVA

By author/editor
Norén, G. Niklas
By organisation
Department of Mathematics
In the same journal
Data mining and knowledge discovery

Search outside of DiVA

GoogleGoogle Scholar
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

Altmetric score

Total: 18 hits
ReferencesLink to record
Permanent link

Direct link