RAPID: Collaborative: Transfer Learning Techniques for Better Response to COVID-19 in the US

点击翻译按钮获取中文摘要

This project will use available data sets for COVID-19 in other countries, and in NYC, Virginia, and Maryland to build compartmental and metapopulation models to quantify the events that transpired there, and what interventions at various stages may have achieved. This will permit gaining control of future situations earlier. The epidemic models developed during this project will lead to innovations in computational epidemiology and enable approaches that mitigate the negative effects of COVID-19 on public health, society, and the economy.Based on publicly available data sets for COVID-19 in other countries, and in NYC, Virginia, and Maryland, the researchers propose to build compartmental and metapopulation models to quantify the events that transpired there, understand the impacts of interventions at various stages, and develop optimal strategies for containing the pandemic. The basic model will subdivide the population into classes according to age, gender, and infectious status; examine the impact of the quarantine that was imposed; and then consider additional strategies that could have been imposed, in particular to reduce contact rates. The project will apply and extend the approach of "transfer learning" to this problem. The research team is well positioned to conduct this research; they have a long history of experience tracking and modeling infectious disease spread (e.g., Ebola, SARS) and are already participating in the CDC forecasting challenge for COVID-19.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.