Format

Send to

Choose Destination
See comment in PubMed Commons below
J Med Syst. 2012 Nov;36 Suppl 1:S25-36. doi: 10.1007/s10916-012-9887-2. Epub 2012 Oct 20.

Summarizing phenotype evolution patterns from report cases.

Author information

1
Department of Electronics and Computer Science, University of Santiago de Compostela, Santiago de Compostela, Spain. maria.taboada@usc.es

Abstract

The need to represent and manage time is implicit in several reasoning processes in medicine. However, this is predominantly obvious in the field of many neurodegenerative disorders, which are characterized by insidious onsets, progressive courses and variable combinations of clinical manifestations in each patient. Therefore, the availability of tools providing high level descriptions of the evolution of phenotype manifestations from patient data is crucial to promote early disease recognition and optimize the diagnostic process. Although many case reports published in the literature do not provide exhaustive temporal information except only key time references, such as disease onset, diagnosis or monitoring time, automatically comparing cases described by temporal clinical manifestation sequences can provide valuable knowledge about the data evolution. In this paper, we demonstrate the usefulness of representing patient case reports of a neurodegenerative disorder as a set of temporal clinical manifestations semantically annotated with a domain phenotype ontology and registered with a time-stamped value. Novel techniques are presented to query and match sets of different manifestation sequences from multiple patient cases, with the aim of automatically inferring phenotype evolution patterns of generic patients for clinical studies. The method was applied to 25 patient report cases from a Spanish study of the domain of cerebrotendinous xanthomatosis. Five evolution patterns were automatically generated to analyze the patient data. The results were evaluated against 49 relevant conclusions drawn from the study, with a precision of 93 % and a recall of 70 %.

PMID:
23085966
DOI:
10.1007/s10916-012-9887-2
[Indexed for MEDLINE]
PubMed Commons home

PubMed Commons

0 comments
How to join PubMed Commons

    Supplemental Content

    Full text links

    Icon for Springer
    Loading ...
    Support Center