Optimizing Non-Pharmaceutical Intervention Strategies Against COVID-19 Using Artificial Intelligence

Abstract

One key task in the early fight against the COVID-19 pandemic was to plan non-pharmaceutical interventions to reduce the spread of the infection while limiting the burden on the society and economy. With more data on the pandemic being generated, it became possible to model both the infection trends and intervention costs, transforming the creation of an intervention plan into a computational optimization problem. This paper proposes a framework developed to help policy-makers plan the best combination of non-pharmaceutical interventions and to change them over time. We developed a hybrid machine-learning epidemiological model to forecast the infection trends, aggregated the socio-economic costs from literature and expert knowledge, and used a multi-objective optimization algorithm to find and evaluate various intervention plans. The framework is modular and easily adjustable to a real-world situation, it is trained and tested with data collected from almost all countries of the world, and its proposed intervention plans generally outperform those used in real life in terms of both the number of infections and intervention costs.

A framework for estimating the socio-economic impacts of non-pharmaceutical interventions against COVID-19

“…After briefly describing the dataset of OxNPIs in the next subsection, we summarize our approach for predicting infections based on active NPIs in the subsequent one. This approach is more fully described in our previous work 8,9 , which also discusses the universal SECs of the OxNPIs.…”

“…We start with the stringency values introduced in our previous study 9 . These values are tuned for WEIRD countries (Western, Educated, Industrialized, Rich, and Democratic, a common grouping in psychological studies) and denoted as universal stringency.…”

“…To predict infections resulting from each OxNPI configuration, we used SEIRD 16 epidemiological model described in our previous work 9 . Daily infections and deaths were retrieved from the OxNPI database for the studied period.…”

“…Yaesoubi et al 7 developed a decision tool to determine when to trigger, continue, or stop physical distancing intervention in order to minimize both deaths from COVID-19 and the intervention duration. Finally, we 8,9 created a tool for finding intervention plans with optimal trade-offs between reducing infections and SEC using multi-objective optimization -and took second place in the "$500K Pandemic Response Challenge" 10 with it.…”

“…The paper begins by using a fixed SEC for each NPI, called the "universal" SEC, taken from our previous study 9 . It compares the cost-effectiveness of the intervention plans produced by our tool, which balances the benefits and SECs of NPIs -when using these universal SECs -with the plans implemented in real life.…”