The turning point and end of an expanding epidemic cannot be precisely forecast
© Brandon Colbert Photography/Moment/Getty Images
The evolution of an expanding epidemic is as unpredictable as next week’s weather.
Many models have been used to try to predict the progression of the COVID-19 pandemic. However, reliable predictions proved unachievable.
A team that included researchers from the Spanish National Research Council has demonstrated that, while models for the spread of infectious diseases can provide accurate short-term predictions, their predictions inevitably grow more inaccurate with time due to small uncertainties in the initial parameters.
The team illustrated this by using a model to predict the spread of COVID-19 in Spain. It accurately replicated the disease’s spread throughout March 2020 but flopped when predicting its subsequent evolution.
Testing their model with made-up observations, the team found that small initial variations were amplified into ever greater uncertainties over time.
The researchers note that this high sensitivity to parameter values will mean that future predictions will only be meaningful within narrow time windows and in probabilistic terms, similar to weather forecasts.
- PNAS 117, 26190–26196 (2020). doi: 10.1073/pnas.2007868117