[Paper Review] From Prediction to Prescription: AI-Based Optimization of Non-Pharmaceutical Interventions for the COVID-19 Pandemic.

Risto Miikkulainen, Olivier Francon+4 more|arXiv (Cornell University)|May 27, 2020

COVID-19 epidemiological studies41 references7 citations

TL;DR

This paper proposes Evolutionary Surrogate-Assisted Prescription (ESP), an AI-driven optimization framework that automatically identifies effective non-pharmaceutical intervention (NPI) strategies for COVID-19 by balancing pandemic control and economic impact. It demonstrates that workplace and school restrictions are most critical, and suggests time-variant, alternating implementation for more reliable outcomes.

ABSTRACT

Several models have been developed to predict how the COVID-19 pandemic spreads, and how it could be contained with non-pharmaceutical interventions (NPIs) such as social distancing restrictions and school and business closures. This paper demonstrates how evolutionary AI could be used to facilitate the next step, i.e. determining most effective intervention strategies automatically. Through evolutionary surrogate-assisted prescription (ESP), it is possible to generate a large number of candidate strategies and evaluate them with predictive models. In principle, strategies can be customized for different countries and locales, and balance the need to contain the pandemic and the need to minimize their economic impact. While still limited by available data, early experiments suggest that workplace and school restrictions are the most important and need to be designed carefully. It also demonstrates that results of lifting restrictions can be unreliable, and suggests creative ways in which restrictions can be implemented softly, e.g. by alternating them over time. As more data becomes available, the approach can be increasingly useful in dealing with COVID-19 as well as possible future pandemics.

Motivation & Objective

• To develop an automated method for identifying optimal non-pharmaceutical interventions (NPIs) during the COVID-19 pandemic.
• To balance the dual objectives of minimizing disease transmission and reducing economic disruption from interventions.
• To enable customizable NPI strategies tailored to specific countries and local contexts.
• To explore the reliability and design of restriction lifting policies using AI-optimized strategies.

Proposed method

• The method employs an evolutionary algorithm to generate a diverse set of candidate NPI strategies.
• It uses predictive epidemiological models as surrogate functions to evaluate the effectiveness of each strategy without full simulation.
• The framework integrates constraints related to public health and economic impact into the optimization process.
• It enables dynamic, time-varying intervention patterns, such as alternating restrictions on schools and workplaces.
• The approach is scalable and adaptable to different locales by adjusting input parameters and data.
• It leverages surrogate modeling to reduce computational cost in evaluating large strategy spaces.

Experimental results

Research questions

• RQ1 What non-pharmaceutical intervention strategies most effectively reduce COVID-19 transmission while minimizing economic impact?
• RQ2 How can intervention strategies be customized for different countries and local contexts?
• RQ3 What are the reliability and long-term outcomes of lifting restrictions after an outbreak?
• RQ4 How can restrictions be implemented in a soft, phased manner to improve sustainability?
• RQ5 What role do workplace and school closures play in controlling pandemic spread?

Key findings

• Workplace and school restrictions emerged as the most impactful NPIs in reducing transmission.
• Lifting restrictions can lead to unreliable outcomes, suggesting risks in abrupt de-escalation.
• Alternating or phased implementation of restrictions improves sustainability and reduces rebound effects.
• AI-optimized strategies can balance public health and economic objectives more effectively than static policies.
• The approach demonstrates feasibility for future pandemics with improved data availability.
• Surrogate-assisted optimization enables efficient exploration of complex intervention strategy spaces.