The United States is waiting with bated breath to see its crucial coronavirus curves – daily cases, hospitalizations, and deaths – flatten, peak and begin to decrease. A number of models have attempted to predict these milestones, most notably a model from the University of Washington’s Institute for Health Metrics and Evaluation: the widely cited IHME model. A new study from the University of Texas at Austin, enabled by resources from the Texas Advanced Computing Center (TACC), has differentiated itself from the IHME model in several ways, coming to its own conclusions about the trajectory of peak COVID-19 deaths in the United States.
While the UT Austin model shares many similarities with the IMHE model, the new model uses only U.S. data and accounts for state-by-state social distancing variables. Employing HPC resources from TACC, the team used anonymized GPS data from tens of millions of cell phones to analyze patterns of social congregation within each state, finding major differences between states that affected each of their COVID-19 curves.
“On average, there seems to be a 3-4 week lag between when someone gets infected with COVID-19 and when they’re at risk of death,” said Lauren Ancel Meyers, professor of integrative biology at UT Austin and head of the UT COVID-19 Modeling Consortium. “That means we’re just beginning to see the life-saving benefits—and the clear signal in the data—of social distancing that began in mid- to late March.”
The UT Austin model and the IHME model both estimate that the U.S. is past its peak daily deaths, which likely occurred around the middle of last week. Their state-by-state projections, however, continue to differ: in DC, for instance, IHME claims that the projected peak occurred five days ago, while UT Austin projects only a 40 percent chance that the peak has already passed, rising to 64 percent in a week.
The UT Austin researchers explain that this is likely due to a third difference between the two models: different methods of quantifying uncertainty. This uncertainty also leads to broad differences in projected daily deaths. When projecting next week’s COVID-19 deaths in DC, the IHME model predicts zero to 16 deaths, while the UT Austin Model predicts one to 34.
“While more uncertain forecasts may be disconcerting, we believe that they reflect the true range of possibilities that could unfold in the weeks ahead,” said James Scott, a professor of information, risk and operations management at UT Austin. “Our model stands on the shoulders of the IHME model, but it corrects critical statistical flaws that led the IHME model to make many projections that, in retrospect, have turned out to be far too optimistic.”
The UT Austin COVID-19 mortality projection model, which is hosted by TACC’s Modeling Consortium portal, can be accessed here.
To read the TACC article reporting on this research, click here.