Simulation and reporting of covid-19
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the efinitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The first challenge when modeling a novel epidemic is that all the reported and inferred data about its dynamics are inevitably affected by some level of uncertainty, resulting in wide ranges and systematic bias. The number of infected cases on any given date is a hidden state in the stochastic process and cannot be observed directly because it is never clear how much time elapses between when infection occurs and when that infection is identified and reported. This gap includes the incubation period and any delay in medical visit, diagnosis, or reporting. To make matters worse, modeling may be influenced by authors' prejudice, interest relationships, or preconceived ideas. Therefore, we argue that scientific research should combine multiple sources of data worldwide rather than be based on a single source, and should treat estimates from global researchers as elastic constraints imposed on models.