The IHME estimates circulating a couple days back, indicated 'true' daily case counts @ 6M+ (currently) for India, off a rather complex model. What didn't sit well with me was the suggestion of an undercount by 28-30x.

Working with these numbers...+ https://twitter.com/UntergrundmannG/status/1385972122791538689
...for over a year now, my sense of the true count ranges in the region of 6-8x, serosurveys + second/third regional waves + undertesting + underreporting all put together.

The WHO paper (shared earlier) on adjustment of TPR seemed to suggest a method to estimate...+
... a reasonable level for the 'true' daily infection counts, using just the testing data and reported daily case counts.

In a pandemic, testing generally scales linearly, while the infection spreads exponentially. Which suggested that the TPR adjustment ... +
... based off the difference in growth rates in daily cases versus daily tests, might provide a heuristic that could be less shocking than the IHME estimates.

Managed to work that out and the numbers seem to line up with my sense of the actual counts. Took a while to ... +
... figure out an approach consistent for all states, given the differences in their reporting and reliability.

The implication for any estimate is the CFR and the fatalities. The Q that's problematic is how long and to what extent can 10x-30x death counts daily be missed.
I don't think people would sit quiet beyond a point.

This (👇) is the table for the major states. Aggregate count off their dailies (estimated as of today) is around 1.2M. If it's worse than this, we're in deep trouble.

ODC: Observed Counts
EDC: Estimated Counts
You can follow @UntergrundmannG.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: