Predicting an uncertain future: improving the utility of computational models during a pandemic

The COVID-19 pandemic has presented policy and decision-makers with a very difficult planning problem. One part of this problem is to anchor decisions in a rapidly evolving and highly uncertain scientific understanding. Computational models have figured prominently in this process.

These models are both powerful integrative devices that structure knowledge and values into actionable information, and epistemically complex tools where decision relevant uncertainties easily become obscured behind model assumptions.

The aim of this project is to (a) assess the use and usefulness of models in the management of COVID-19 in Sweden, and (b) develop guidelines and best-practices for how to better integrate models and modelling in decision making in public health going forward. Our approach combines epidemiology, mathematical modelling, medical informatics, and philosophy of science.

The work decomposes into three parts: the 1st consists of evaluating and analysing the models used to support decision-making during the COVID-19 pandemic focusing both on the specifics of the models and the experiences of the decision makers. The 2nd uses these insights to develop a framework that can guide similar efforts in the future.

The 3rd synthesises the gained knowledge into suggested best-practices.

This project will lead to better preparedness for the next pandemic, and improved practises for integrating computational methods, including AI and algorithms, into the decision-making process.