Tuesday, May 5, 2020

Garbage at Healthdata.org

The IMHE models at Healthdata.org are just garbage.  If you looked at the US and Texas yesterday and then today they are completely different.  Yesterday we had lots of little small ups and downs in the curves for the prior couple of weeks.  Today, those have disappeared (the curves became relatively smooth) and the numbers nearly doubled.  They even changed the types of data they were reporting on.  Any curve that used to approach an asymptote (they leveled off), now keeps heading towards infinity.

This is all just crap.  In previous blogs, I've told you why the models are bad.  I also described how a simple back of the envelope assessment would show that infections would continue to increase until we had herd immunity or a vaccine.  I guess the virus could also mutate into an innocuous strain.

All of the IMHE models used to have a hump and then went down.  Now, they have a big, big hump and barely go down.

My conclusion is that IMHE simply uses assumptions about mitigation efficiency from lock downs and social distancing as the PRIMARY variables in their models.  That is, you can change variables to change the model's output, and the variable that affects the output the most is what I am calling the primary variable.

They are apparently not using past data to validate their models and do smaller and smaller refinements.  They are jumping an effective R from 1.1 to 2.0 (I made those numbers up) because they see some states starting to open.  That is, they made a radical change in a primary variable.

The issue is that it has been less than a week since openings began.  It takes 2 to 14 days to develop symptoms and up to another two weeks for a person to die.  They have absolutely NO data on what state openings are going to do to the numbers.  They are making guesses.

Have you been surprised at how accurate the models have been in the past?  Me neither; they've been horribly inaccurate and a failure at prediction.  Why would you want to believe their new guesses?

I hate to say it, maybe they put their best modelers on this (ha ha).  But it sure looks like more fear mongering to me without a sound basis for any kind of reasonable estimate.

So once again, this is all just garbage.

No comments:

Post a Comment