Human Epidemiology and RespOnse to SARS-CoV-2 (HEROS): Objectives, Design and Enrollment Results of a 12-City Remote Observational Surveillance Study of Households with Children using Direct-to-Participant Methods (preprint)

The Human Epidemiology and Response to SARS-CoV-2 (HEROS) is a prospective multi-city 6-month incidence study which was conducted from May 2020-February 2021. The objectives were to identify risk factors for SARS-CoV-2 infection and household transmission among children and people with asthma and allergic diseases, and to use the host nasal transcriptome sampled longitudinally to understand infection risk and sequelae at the molecular level. To overcome challenges of clinical study implementation due to the coronavirus pandemic, this surveillance study used direct-to-participant methods to remotely enroll and prospectively follow eligible children who are participants in other NIH-funded pediatric research studies and their household members. Households participated in weekly surveys and biweekly nasal sampling regardless of symptoms. The aim of this report is to widely share the methods and study instruments and to describe the rationale, design, execution, logistics and characteristics of a large, observational, household-based, remote cohort study of SARS-CoV-2 infection and transmission in households with children. The study enrolled a total of 5,598 individuals, including 1,913 principal participants (children), 1,913 primary caregivers, 729 secondary caregivers and 1,043 other household children. This study was successfully implemented without necessitating any in-person research visits and provides an approach for rapid execution of clinical research.

Federated Learning used for predicting outcomes in SARS-COV-2 patients (preprint)

‘Federated Learning’ (FL) is a method to train Artificial Intelligence (AI) models with data from multiple sources while maintaining anonymity of the data thus removing many barriers to data sharing. During the SARS-COV-2 pandemic, 20 institutes collaborated on a healthcare FL study to predict future oxygen requirements of infected patients using inputs of vital signs, laboratory data, and chest x-rays, constituting the “EXAM” (EMR CXR AI Model) model. EXAM achieved an average Area Under the Curve (AUC) of over 0.92, an average improvement of 16%, and a 38% increase in generalisability over local models. The FL paradigm was successfully applied to facilitate a rapid data science collaboration without data exchange, resulting in a model that generalised across heterogeneous, unharmonized datasets. This provided the broader healthcare community with a validated model to respond to COVID-19 challenges, as well as set the stage for broader use of FL in healthcare.

Working from a Distance: Productivity Dispersion and Labor Reallocation (preprint)

Following the shocks of the COVID-19 pandemic, the economy may be significantly changed relative to the pre-pandemic world. One critical shift induced by the COVID-19 pandemic is a need for physical distance (at least 6 feet apart) between workers and customers. In this study, we examine the impacts of social distancing in the workplace on employment and productivity across industries. Using our constructed measure of adaptability to social distancing, we empirically find that industries that are more adaptive to social distancing had less decline in employment and productivity during the pandemic. Using this empirical evidence, our model predicts that employment and productivity dispersion would induce labor reallocation across sectors, while imperfect labor mobility may result in a long road to economic recovery.