Evaluating the effect of city lock-down on controlling COVID-19 propagation through deep learning and network science models

TL;DR

This study uses deep learning and network science to evaluate how city lock-downs affected COVID-19 spread, revealing alternative measures that could achieve similar containment without full lockdowns.

Contribution

It introduces a robust NP-Net-SIR model trained with supervised learning to analyze COVID-19 data, accounting for poor data quality and exploring lock-down alternatives.

Findings

Lock-downs are effective but not the only containment measure.

Certain non-lock-down measures can achieve similar infection control.

Model provides guidelines for flexible COVID-19 containment strategies.

Abstract

The special epistemic characteristics of the COVID-19, such as the long incubation period and the infection through asymptomatic cases, put severe challenge to the containment of its outbreak. By the end of March 2020, China has successfully controlled the within-spreading of COVID-19 at a high cost of locking down most of its major cities, including the epicenter, Wuhan. Since the low accuracy of outbreak data before the mid of Feb. 2020 forms a major technical concern on those studies based on statistic inference from the early outbreak. We apply the supervised learning techniques to identify and train NP-Net-SIR model which turns out robust under poor data quality condition. By the trained model parameters, we analyze the connection between population flow and the cross-regional infection connection strength, based on which a set of counterfactual analysis is carried out to study the necessity of lock-down and substitutability between lock-down and the other containment measures. Our findings support the existence of non-lock-down-typed measures that can reach the same containment consequence as the lock-down, and provide useful guideline for the design of a more flexible containment strategy.