Academic

Temporal Data Mining in Health

CSIRO's Division of Mathematical & Information Sciences (CMIS) is seeking to appoint a high-calibre Postdoctoral Research Fellow to perform research in temporal data mining. New methods and algorithms will be developed for the automatic identification of interesting patterns from large longitudinal health data sets. The position is for a period of 3 years, starting in 2002. The position is located with CSIRO Data Mining on the campus of the Australian National University, Canberra, Australia.

A focus of the CSIRO Data Mining group is the application of data mining technologies in health services research. A recent example is de-identified hospital data and de-identified Medicare and Pharmaceutical data from the Commonwealth Department of Health and Ageing which have been linked to provide insights into the delivery of health care. As with many such administrative datasets this linked dataset has a strong temporal component with patient histories available over several years. Financial/banking data, taxation data, and supermarket basket data are other examples of administrative datasets with a strong temporal component. The temporal relationship between items is an important aspect of this data.

Working in the context of the large linked and temporal datasets available the postdoctoral fellow will work with and under the guidance of data mining researchers. Specifically, an approach to capture the concept of surprise in the context of the longitudinal data available will be developed and deployed to identify, automatically, `surprising' discoveries.

The successful applicant will have a recent Ph.D in computer science, statistical computing, or related disciplines. Demonstrable research skills in data mining preferably in temporal data mining. Good programming skills are also necessary.

The Position Description, Selection Criteria, and Application information can be found at http://datamining.csiro.au/edm/postdoc.html