10 April 2020

Checking the Lock Down's Effect

The lock down model is confusing at this point:1

Red= Should have been, Purple=Infected Prior, Green=New Infected
Prior infected should have increased for five days after lock down (avg incubation period) and then dropped off. I didn't make the drop off entirely linear, but had them drop off more steeply closer to the twelfth day when 97% of prior infected should have shown symptoms (days and percentages from here).  The new cases I put in start with 16 and add 16 more to the new cases each day which was an average number I came up with for the increase in new infections from 10 March onward. I could have used the median of 12, and think that might be the correct number to use, but went with 16 out of an abundance of caution. I did make the original line upward for prior infected add 12 more to the last number of newly infected per day because that was the average from 10 March before that date.

YES, before someone tells me that I don't know what I'm doing and should have used "the advanced confusion model of non-understandable mathometrics", I concede that there is no way I could make a perfect model. It's flawed, but close enough to further intelligent discussion.

And, here's the rub: the numbers end up about where I projected they would, but that feels an awful lot like an anomaly. I suspect that if the red and green lines continue to project upon their merged path they won't correspond with future data. However, that's at best an educated guess. I'll have to recheck the numbers in a few days.

Interestingly, the actual numbers didn't follow what they should have. They should have gone higher and then dropped back down. The first 5 days go up until the weekend and after the weekend it's at a higher level. Then it flattens. This could mean the post-weekend numbers are inflated and should really be lower (some cases should shift to Saturday and Sunday) and, if accurately measured Monday and Tuesday would have smaller numbers and reflect a slight growth for the weekdays. No matter whether that is true or not, it seems to indicate that the green line should be flatter, but that would mean the final result would not be consistent with the numbers at the end. Is the green line subject to the nigh unto mythical "exponential infection rate?" If so, numbers should continue to rise steeply into the next week. Again, something that needs further future data to develop.

The biggest flaw in all of this is the fact that increased testing has become available. This results in its own, unmeasurable bump in the numbers. I don't know how much to factor in for that in. For goodness sake, some places have been offering drive through testing and the experience where I'm at is that it's moved from impossible to get tests to available if needed. So, it almost goes without saying that there's a significant bump except I have no way of knowing what it is and factoring it in.


1  This was moved from the back end of the previous post because the two are about different things.

No comments: