We model the growth of scientific literature related to COVID-19 and forecast the expected growth from 1 June 2021. Considering the significant scientific and financial efforts made by the research community to find solutions to end the COVID-19 pandemic, an unprecedented volume of scientific outputs is being produced. This questions the capacity of scientists, politicians and citizens to maintain infrastructure, digest content and take scientifically informed decisions. A crucial aspect is to make predictions to prepare for such a large corpus of scientific literature. Here we base our predictions on the Autoregressive Integrated Moving Average (ARIMA) and exponential smoothing models using the Dimensions database. This source has the particularity of including in the metadata information on the date in which papers were indexed. We present global predictions, plus predictions in three specific settings: by type of access (Open Access), by domain-specific repository (SSRN and MedRxiv) and by several research fields. We conclude by discussing our findings.

We present a forecasting analysis on the growth of scientific literature related to COVID-19 expected for 2021. Considering the paramount scientific and financial efforts made by the research community to find solutions to end the COVID-19 pandemic, an unprecedented volume of scientific outputs is being produced. This questions the capacity of scientists, politicians and citizens to maintain infrastructure, digest content and take scientifically informed decisions. A crucial aspect is to make predictions to prepare for such a large corpus of scientific literature. Here we base our predictions on the ARIMA model and use two different data sources: the Dimensions and World Health Organization COVID-19 databases. These two sources have the particularity of including in the metadata information the date in which papers were indexed. We present global predictions, plus predictions in three specific settings: type of access (Open Access), NLM source (PubMed and PMC), and domain-specific repository (SSRN and MedRxiv). We conclude by discussing our findings.