So there are numerous headlines about increased COVID propagation this month. It is primarily coming from Epidemiologic models with many assumptions. My modeling work was purely mathematical – fitting non-linear functions (in my case Logistic functions) to time series data.
Disease models use many assumptions about transmission rates, mobility, transition probabilities through various states of disease, etc.. The IHME model was more of a pure mathematical model until this latest revision, where they moved to a hybrid model.
I believe they painted themselves into a corner when they declared some weeks ago that no relaxation of restrictions was safe until a state had less than 1 COVID case per million of population.
No state has (or will likely) meet that criteria. We likely don’t meet that criteria with flu even this late in the season. So with states opening up now, they really had to project a worsening of results.
, the problem with the assumption driven models is that they have very wide confidence intervals – a few tweaks in assumptions and they produce very different results. What we see happening now is that some modelers are ASSUMING that relaxing restrictions must have a significant negative effect. I’m not so sure.
The disease was already slowing down in late March before most of the lockdowns could have any effect. The lockdowns didn’t change much, and I expect releasing them won’t change much either. Why? Simply because the lockdowns are not the most significant motivator of human behavior. People started self-preservation behavior long before they were told to, and they will adapt to the great opening by finding new and continuing ways to be cautious.
In any event, we can speculate about growth in COVID over the month of May, but there is no evidence of it yet. The data are still in decline.