Online Program

Examining Leading Indicators for COVID-19 Incidence Case Growth (309734)

*Vishnu Shankar, Stanford University 
Kate Harwood, Google 
Sangwon Hyun, University of Southern California 
Ben Smith, Google 
Ryan Tibshirani, Carnegie Mellon University 

Keywords: COVID-19, Leading Indicators, Disease Forecasting, Public Health

Throughout the COVID-19 pandemic, the U.S. has seen multiple “waves” of increases in case incidence numbers, occurring at different times in different areas of the country, and with even greater variation on the state and county level. It has been difficult not only for public health officials and other policy makers to anticipate when case incidences might rise in their area, but also for researchers who are forecasting the pandemic to accurately predict future case numbers. Part of the challenge in guiding decision making has been understanding how to leverage the combination of mobility, public health, and survey signals to capture the complex dynamics and transmission of COVID-19.

To address this challenge, we analyze several signals - processed and made publicly available by the Delphi Research Group - as potential leading indicators of case growth on the county level. One example is the doctors visits signal, which is the estimated proportion of outpatient doctor visits primarily about COVID-related symptoms, based on data from Delphi’s health system partners. The goal is to provide greater insight into the value of these signals in predicting future increases in cases.

Our analysis is two-fold. First, we construct a method based on smoothed numerical derivatives to identify points where a signal begins to rise significantly. Using this method, we are able to mark where indicator signals and cases rise significantly throughout all counties over the course of the pandemic. Next, we analyze how well rises in indicators precede rises in cases using several carefully defined performance measures.

Among other findings, our results show that most leading signals lead substantial case rises by a week on average, and the combination of signals can provide independent value in anticipating case rises. We believe that our results and reusable methodology can be useful for assessing the value of other potential leading indicators in infectious disease spikes.