Two bioinformatics experts have used AI to accurately estimate regional COVID infection rates, aka seroprevalence, in all 50 states and 50 hard-hit countries.
Their computational method beats antibody testing at the task while showing the testing’s primary purpose, seroprevalence surveys of small sample groups, badly missing the mark.
In fact the method shows U.S. infections almost three times greater than the numbers reported based on the surveys.
Jungsik Noh, PhD, and Gaudenz Danuser, PhD, both of the University of Texas Southwestern Medical Center, describe their work in a study posted Feb. 8 in Plos One.
Noh and Danuser developed their model using established epidemiological parameters as well as publicly available data on confirmed cases and deaths.
Comparing their algorithm’s estimates against those of the corresponding seroprevalence surveys, the team found the surveys yielding “severe under-ascertainment” of COVID cases across the U.S. and in the 50 other countries they analyzed.
In half the 50 countries—all of which had high infection rates during the study period—the algorithm’s estimates of actual cumulative cases were up to 20 times greater than the cases confirmed by antibody testing.
In their discussion, Noh and Danuser emphasize that, unlike seroprevalence surveys, their computational approach can supply useful infection-rate estimates and update them every day.
“More importantly, our framework estimates the actual fraction of currently infected people in each region,” they add. “The estimated number of current infections can serve as an initial target in planning effective contact tracing. Since the developed pipeline requires simple input, [the framework] is widely applicable to more granular analyses of specific regions or communities, for which the number of confirmed cases and deaths are being tracked.”
Further, Noh and Danuser write, antibody testing’s resulting “severe under-ascertainment” of COVID cases has shrouded the reach and severity of the pandemic worldwide.
“Given that the confirmed cases only capture the tip of the iceberg in the middle of the pandemic,” they add, “the estimated sizes of current infections in this study provide crucial information to determine the regional severity of COVID-19 that can be misguided by the confirmed cases.”
Full study here.