Bless you! New pollen surveillance system to reduce respiratory disease risk
In November 2016, a high grass pollen count, stormy weather and strong winds conspired to create a deadly thunderstorm asthma event in Melbourne. Nine people died from severe asthma attacks and hundreds more overwhelmed ambulance services and hospital emergency departments.
According to UTS Chancellor’s Postdoctoral Fellow Qiaoyun Xie, a national pollen forecasting system could have given hospitals, ambulance services and allergy sufferers precious time to prepare for the danger that lay ahead. But, while such systems exist in other countries, Australia is lagging behind.
In response, Xie is using her combined expertise in science and engineering, including machine learning and remote sensing, to embark on a four-year project to establish a national pollen surveillance system in Australia.
The system will combine meteorological and satellite information with data from time-lapse cameras and pollen collection traps in New South Wales, Queensland and Victoria to produce long- and short-range grass pollen forecasts.
“I will use remote sensing geospatial techniques to first track the grass growing conditions, and then use historical satellite data going back to the last century to train my deep learning model,” says Xie, who is based in the UTS School of Life Sciences.
“Both from the satellite data and the time-lapse camera, I will extract the grass canopy green-ness information, which is an approximation of grass growth conditions. Having that year by year together with the meteorology data, I will train the model to predict nationwide grass growth and pollen concentration for the current season.
“I already have the pollen count of the past month, so I can compare my forecast data with the real pollen count to see the accuracy of my model.”
Long-range forecasts can provide a picture of what to expect in terms of pollen concentrations up to three months in advance. By contrast, the short-range forecast will provide daily and weekly pollen count predictions, similar to the way the Bureau of Meteorology provides daily weather forecasts that are available in the public domain.
“For people who have that allergy to pollen, they can weigh this information and see whether they need some medicine at hand or not, or whether they should avoid outdoor activities or not,” Xie says.
As well as making a valuable contribution to public health, the work could also help governments, health insurers and communities prepare for the health challenges associated with high pollen counts, as well as helping pharmaceutical companies produce higher volumes of anti-allergy medication during periods of allergy threat.
The work is a partnership with researchers at Macquarie University and the Queensland University of Technology.