Abstract for presentation at Spatial Sciences Institute International Biennial Conference

Development of algorithms for the early detection of disease outbreaks in Australia

  • Serryn Eagleson, Australian Biosecurity CRC, Curtin University of Technology, Australia
  • Rochelle Watkins, Australian Biosecurity CRC, Curtin University of Technology, Australia
  • Bert Veenendaal, Australian Biosecurity CRC, Curtin University of Technology
  • Graeme Wright, Australian Biosecurity CRC; Curtin University of Technology, Australia
  • Aileen Plant, Australian Biosecurity CRC; Curtin University of Technology, Australia

    • As tourism and trade increase world-wide so does the risk for the transmission of disease. Improved capacity for early and rapid detection of disease outbreaks is therefore becoming more critical if health professionals are to provide an effective response.
    The purpose of this research is to develop tools that will utilise existing surveillance data to assist disease control in Australia. The problems associated with detecting disease outbreaks are complex. The number of cases indicating the presence of an outbreak can often vary according to the disease, population, scale and level of aggregation of the data. In Australia there are approximately sixty different notifiable diseases aggregated to residential postcode.
    Within the spatial sciences discipline a number of spatial analysis techniques have been developed to facilitate the identification of disease clusters. However, very few of these have the capacity to analyze spatial statistics through time. This is important as the spread of disease is a dynamic process, and the pattern at a fixed point in time is not very informative about the way the pattern has changed through time. The challenge is to identify areas in which spatial and/or temporal anomalies exist.
    To address this problem the authors have developed a number of spatiotemporal algorithms aimed at detecting disease outbreaks before they become widespread. The algorithms have been developed using Local Area Indicator Statistics (LISA) such as Local Morans I and Tangos MEET statistic and are used to monitor changing values of disease occurrences over time.
    The algorithms are embedded within a GIS prototype. The prototype has been designed using two modules. Module 1 controls the configuration and operation of the algorithms. Module 2 facilitates data exploration and visualisation. Using this module the user is able to select the disease and time period of interest. In combination these modules are aimed at providing an environment in which the analyst can identify abnormal clusters as well as highlight patterns of disease that were previously unknown.

    Conference Organiser - ICMS Pty Ltd