Epidemics can isolate people from one another, negatively affect the health of people across the world and cause economic declines that take countries months to years to recover from. Avoiding these epidemic impacts can be essential to preserving human wellbeing, but it remains extremely difficult when there is little understanding of how the disease will affect society or when and how to effectively use strategies to lower transmission of the disease.
One strategy for avoiding these pronounced negative health, social and economic impacts is through the development of models that are capable of predicting epidemic waves and issuing warnings when cases are expected to increase in the near future. These models can help public officials determine when and where to deploy intervention methods to curb the spread of the disease.
Using statistics and reported case data, a research team, including CEID’s Pejman Rohani, developed one such model and utilized it to analyze current COVID-19 case numbers and trends to determine if a future spike in cases was probable. The model issued a warning to expect an increase in cases if the change in case numbers exceeded a certain threshold and the number of reported cases was more than the average number of cases reported the previous week.
Both the size of the rolling window for case numbers and the threshold for the relative change in case numbers can be altered based on the desired accuracy and purpose of the model. Though the research team used the model to assess COVID-19, the model could potentially be used for other similar diseases.
To test the accuracy of their model, they applied it to real COVID-19 case number data from Italy and New York between January 22, 2020, and April 13, 2021. The model was able to correctly issue a warning in Italy 82% of the time while it only inaccurately issued a warning 9% of the time. For New York, it accurately issued warnings 55% of the time, and it only issued warnings unnecessarily 9% of the time. This indicates a promising ability to accurately predict increases in case numbers.
The research team hopes that this model will be integrated into the current public health sector and One Health initiatives to be used to lower transmission rates in future epidemics. Tools like this model are essential for public officials to determine when and where to utilize intervention methods to disrupt pathogen spread.
By Amanda Budd