Alice Richardson (University of Canberra)
Enhanced prediction of infectious diseases though routine pathology data
Pathology laboratories around Australia generate large amounts of data based on the requests for tests placed by medical practitioners. The resulting databases represent a largely untapped resource of observational data that has the potential to improve laboratory diagnosis of disease and reduce redundancy in testing through the objective identification of associations between test results.
In this talk I’ll describe some recent attempts to make use of this data to these ends. The initial focus will be on enhanced laboratory diagnosis of infectious diseases such as the lung condition Chlamydia pneumoniae, and Hepatitis B and C. I’ll describe the nature of the available data and consider its quality in terms of variability and missingness. I’ll also describe results of experiments to determine the best combinations of data pre-processing and statistical analysis methods to enhance prediction.
Finally, I’ll consider the use of routine pathology data and other clinical information for the diagnosis of more complex conditions such as Chronic Fatigue Syndrome.