?:abstract
|
-
A two-parameter, human behavior Covid-19 infection growth model predicts total infections between -4.2% (overprediction) and 4.5% (underprediction) of actual infections from July 27, 2020 to September 30, 2020 for 10 US States (NY, WA, GA, IL, MN, FL, OH, MI, CA, NC). During that time, total Covid-19 infections for 9 of the 10 modeled US States grew by 60% (MI) to 95% (MN). Only NY limited Covid-19 infection growth with an 11% increase from July 27 to September 30, 2020. September is a month with contraposing effects of increased social interaction (eg, physical school openings) and outdoor temperatures decreasing to the 50F (10C) to 70F (21C) range in which outdoor activities and building ventilation are beneficially increased. All State infection predictions except GA, FL and CA predictions through September 30 are bounded by four prediction scenarios (no school with outdoor temperature effect, no school with no outdoor temperature effect, school with temperature effect, school with no temperature effect). GA, FL and CA continued along a path slightly below the linear infection growth boundary separating infection growth and decay, resulting in overprediction of infection growth over the two month simulation period(-3.1% for GA, -1.9% for FL, and -4.5% for CA). Three eastern States (NY, NC, and GA) are most accurately represented by models that assume no significant change in social interactions coupled with minor outdoor temperature effects. Four midwestern States (IL, MI, MN, OH) are most accurately modeled with minor outdoor temperature effects due to a delayed decrease in average outdoor temperatures in the Midwest. The remaining three States (WA, FL, and CA) are also in good agreement with the model but with differing weather condition and social interaction impacts. Overall, model predictions continue to support the basic premise that human behavior in the US oscillates across a linear infection growth boundary that divides accelerated infection growth and decaying infection transmission.
|