Using semantic and graph similarity methods to predict heart or kidney abnormalities
Keywords
Loading...
Authors
Issue Date
2022-01-30
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
There are many symptoms, which cannot be visually spotted. Imaging is
needed to spot these symptoms, which is time intensive. Two examples of
these symptoms are heart abnormalities and kidney abnormalities. These
can be predicted with the use of graph and semantic similarity methods.
Patients are represented as HPO graphs with symptoms being HPO nodes
in the graph. Graph and semantic similarity can be used to compare patients
with each other in order to make a classification based on the most
similar patients. Resnik and Lin, two semantic similarity methods, have
been compared to the maximum common sub-graph method, which is a
graph similarity method. Overall the maximum common sub-graph method
yields the best performance and has an AUC of 0.89. This can already be
used with this data to determine the priority of the order in which patients
undergo imaging.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen