Department     College of Asia Pacific Studies
   Position   Professor
Language English
Publication Date 2011/08
Type Bulletin of university, Research institution
Peer Review Peer reviewed
Title Hybrid Approach to Data Mining Radiological Medical Records
Contribution Type Joint Work
Journal Geoinformatics Research Center Publications
Author and coauthor William Claster, Subana Shanmuganathan and Nader Ghotbi
Details Here we develop a hybrid
methodology for the analysis of medical records that
consist of clinicians’ notes (requesting for a scan) using
statistical tools and Kohonen’s self-organizing map
(SOM) text mining techniques to look for heretofore
invisible patterns to save patients from unwarranted
scans. The medical data derive from patients’ radiology
department records where CT (Computed Tomography)
scanning was used as part of a diagnostic exploration.
The records are from the digital records of about 700
pediatric patients who underwent CT scanning (single
and multiple) through a one-year period in 2004 at the
Nagasaki University Medical Hospital in Japan. This
approach led to a model based on SOM clusters and
statistical analysis which allow for the prediction of when
a particular medical screening procedure may be
unnecessary. The procedure involves CT scans of
Note: This is important because radiation at levels
ordinarily used for CT scanning may pose significant
health risks especially to children.