Conversation

Jason Smith
@infotranecon
The improvement in care in the aggregate data appears to have stopped. Looking at the ratio of national deaths to hospitalizations we can clearly see the improvement up until the summer. However, since then the ratio has stayed relatively steady around 2-3%.
Image
Quote Tweet
The Atlantic
@TheAtlantic
Many Americans are under the impression that no matter how high cases surge in the coming weeks, deaths will not reach the heights of the spring. The truth is more grim, write @alexismadrigal and @whet. on.theatln.tc/WEYG9Zf
1
5
Jason Smith
@infotranecon
If we use this to estimate deaths based on hospitalizations forecast, we are looking at anywhere between 2000 and 6000 deaths per day in January.
Image
1
5
Jason Smith
@infotranecon
Replying to
@infotranecon
@jasonbaumgartne
and
@ZakDavid
You can detect the beginnings of the surges using a simple threshold detector. The estimate of a new surge is wildly uncertain at first, but eventually calms down as we pass the the most rapid rate of change. Animation here:
Quote Tweet
Jason Smith
@infotranecon
Animation showing detection of each new surge pulse (vertical gray lines) and the best estimate of the path (blue lines). You can see the instability around the initial phase of the a surge — an indicator the surge is just going to get worse.
Show this thread
Embedded video
GIF
1
2