|
|
|
|
|
|
|
|
|
| CSIRO | SOLVE | Issue 3 - May 05 | |
|
|
|
ARTICLE
DATA MINING:
Health Databases get a Bioterrorism Check-Up By JIM BUCKELL
The extent of any health crisis caused by bioterrorism or 'superbugs' will be influenced primarily by how quickly authorities detect the presence of the disease agent.Consequently, health authorities are looking for early-warning strategies - the basis of which may exist already in health industry databases. The concept being explored is that any new set of disease symptoms reported by doctors or hospitals might become more quickly apparent if there was a way to make anything unusual stand out against the normal backdrop of epidemics like flu, measles and other 'routine' infections. To this end, researchers in Australia are probing databases to find information that can be used to create this 'screen'. Statisticians led by Dr Christine O'Keefe and Dr Ross Sparks, of CSIRO Mathematical and Information Sciences, are working on complex mathematical models built around data held in medical service computers. Specifically, data from the emergency departments of eight Sydney hospitals is being examined to identify outbreaks of influenza and other acute-onset illnesses. Epidemiologists Dr Tim Churches and Dr David Muscatello of NSW Health, who are coordinating the data mining, say the detection of possible bioterrorism agents is complicated because in the early stages many of them may produce symptoms that are indistinguishable from common respiratory diseases. So the task is to ascertain whether or not a spike in the number of people with particular sets of symptoms indicates the presence of a new threat to public health, such as a toxic agent or virulent strain of flu. The statistical methods being developed by the CSIRO team will allow the constant streams of health data to be analysed by computer on a near real-time basis. This means that potentially harmful outbreaks of disease can be detected earlier and emergency strategies put in place to prevent its spread. The same technology is being used in another project, with the Federal Department of Health and Ageing, to measure adverse side effects of medication. The team has developed a prototype software program called iHealth Explorer that mines the data on medical treatment held by hospitals. The program links this information with data held by the Pharmaceutical Benefits Scheme to locate drug reactions that cause harmful side effects. Researchers say the two projects show that applications for sophisticated data mining could be boundless.
For further information contact: |
IN THIS ISSUE
|
| Home | About Us | eSubscribe | Links | |
|
|
||
|
|
|
Back to main