Health data mining with Medicare data.
Rohan Baxter
CSIRO-MIS
Tuesday 12 June at 11am
Abstract
CSIRO is helping the Commonwealth Department
of Health and Aged Care to discover useful
public health information from their massive collections of medical
data. By using advanced data-mining techniques to analyse Medicare
and PBS (Pharmaceutical Benefits Scheme) data, the team discovered
what type of patients were most likely to miss out on full care
and the range of diagnostic tests recommended for diabetes patients.
They found that patients in rural areas and elderly patients were
most likely to not receive full care.
The information allows the Department to assess the quality of
the Australian health care system and to plan cost effective public
health campaigns, or provide information or training to doctors,
which target the areas of greatest need. Besides quality, the
technique is being applied to investigate other major public health
issues such as effectiveness, access and financing. The
technique allows rapid testing of many hypotheses on massively
linked data sets. It also employs statistical algorithms for identifying
similar patterns of care.
This seminar describes the methodology employed and the trials
and tribulations behind the scenes of this health data mining
project.
Short resume
Rohan Baxter is a Research Scientist in the Enterprise Data Mining
at CSIRO Mathematical and Information Sciences Division, based
in Canberra. In 1997, he founded Ultimode Inc., a data mining
research and consultancy firm, in Berkeley, California. He obtained
funding as a NSF and NASA SBIR principal investigator. Ultimode
developed a recommendation engine based on patented dyanamic segmentation
algorithms to provide internet personalisation and customer profiling
services to e-commerce companies. He has a PhD in Computer Science
(Machine Learning) from the Department of Computer Science, Monash
University in Melbourne, Australia.
Back to HAIL Home Page