This study aims to develop a curve-fitting approach for long-term COVID-19 mortality projections and evaluate its effectiveness as a scalable, data-driven tool for pandemic forecasting. The basic characteristics of a dynamic curve-fitting approach capable of generating long-term projections are described. To demonstrate its utility, the model was retrospectively applied using mortality data from the start of the pandemic, January to June 2020 (6-month data), to project into the period between June 2020 and April 2021 (11-month projections). For scenarios with the best fit, the difference between observed and projected total deaths varied in the projection period between 7.7% and 28.2%. When the COVID-19 pandemic started in early 2020, there was lack of understanding regarding its long-term impact. Available mathematical models were complex and typically provided short- and mid-term projections. The approach described generates long-term projections that are relatively easy to implement and can be enhanced to include other parameters such as vaccine impact or virus variants. The method could prove to be a valuable tool during a future pandemic.
Kafatos et al. (Wed,) studied this question.