Though many of us do (allegedly) take our smart phones with us to the bathroom, most would never think to link the algorithms at our fingertips – and what’s going on in the plumbing beneath – with staffing at the local hospital.
Yet according to an Australian-led study published in the latest Nature Communications, scientists have recently been able to predict new, weekly COVID-related hospital admissions in over 150 US counties just by analysing wastewater.
While wastewater-based epidemiology (WBE) was already considered an efficient way of providing unbiased estimations of the level of COVID within a community, few studies have reported the association between CRNA (primary sludge) in wastewater with hospitalisations.
Co-lead author, Dr Xuan Li from the Sydney’s University of Technology (UTS) believed that this was the first time that researchers had endeavoured to create surveillance models for forecasting hospital admissions using the approach.
“We evaluated the feasibility of using WBE to predict COVID-induced weekly new hospitalisations in 159 counties across 45 states, covering a population of nearly 100 million,” Dr Li said.
“Using county-level weekly wastewater surveillance data over 20 months, WBE-based models accurately predicted the county-level weekly new admissions, allowing a preparation window of 1-4 weeks.”
Specifically, periodically updated WBE-based models showed good accuracy and transferability, with a mean absolute error within 4-6 patients/100k population for upcoming weekly new hospitalisation rates.
Lead author, UTS’s Professor Qilin Wang, said the research showed wastewater surveillance combined with AI-based modelling could be a cost-effective early warning system, allowing public health officials to better prepare for and manage pandemic waves, and efficiently allocate limited healthcare resources.
“Wastewater monitoring is already conducted in many countries, but it is limited to showing whether COVID is present in a region, as well as a rough estimation of whether the burden is increasing or decreasing,” he explained.
“Variables that can influence hospital admissions include changing behaviour due to public policies, vaccination rates, holidays and weather; and we used artificial intelligence to pick up patterns and changes in the data and learn from this to increase the accuracy of predictions.”
“Current prediction methods are based on COVID laboratory testing, or self-testing and reporting, however this does not pick up asymptomatic cases, and [we] are moving away from rigorous testing requirements,” Dr Li said.
“Yet the number of Australians in hospital with COVID continues to fluctuate, and rapid increases in patient numbers can stress frontline healthcare capacity and increase fatality rates.”
Dr Li was recently awarded a two-year grant from the Australian Academy of Science WH Gladstones Population and Environment Fund to develop an Australian-based WBE prediction model.
She hopes to extend her research to include other infectious diseases that can be detected through WBE, including food-borne pathogens such as salmonella and E-coli, and viruses such as flu, norovirus, and hepatitis A.