A University of Massachusetts model sees the possibility of the coronavirus case rate declining steadily in the coming weeks.
The ensemble model, which combines a number of respected forecasts, says that by Jan. 23, the state could be tallying 26,350 confirmed and probable cases a week. That’s down from 30,332 for the week ending on Saturday.
Researchers noted that the actual numbers on Jan. 23 could range much higher or lower -- from about 12,000 to about 46,000.
The UMass model tended to underestimate the state’s alarming second surge during the fall. But the model caught up, and in the past few weeks it tended to overestimate the actual numbers as the surge flagged slightly.
The latest forecast offers a tantalizing suggestion of a decline in cases in the weeks ahead -- at the same time that officials have voiced dire warnings about a possible spike in cases following the winter holidays.
While cases may decline, the total number of deaths will keep climbing. The model predicts the death toll in the state could rise to around 13,654 by Jan. 23. As of Monday, the number of confirmed and probable COVID-19 deaths in Massachusetts stood at 12,158.
Nationally, the model sees the death toll by that date, three days after President Donald Trump leaves office, rising to stunning heights -- nearly 404,000.
The projection comes from a lab headed by UMass Amherst associate professor Nicholas Reich. The lab collects models from numerous researchers and develops a combined forecast that is intended to reflect their collective wisdom. Reich’s lab only creates the forecast for a four-week window ahead because researchers believe forecasts aren’t reliable enough after that.
Reich’s lab posts its national- and state-level data every week at the Reich Lab COVID-19 Forecast Hub. The lab, already an Influenza Forecasting Center of Excellence, collaborates with the US Centers for Disease Control and Prevention on coronavirus predictions. The lab feeds the data it has collected and its ensemble forecast to the agency, which posts the data on its own website.
Martin Finucane can be reached at email@example.com.