The puzzle of society, civilization,
evolution and systems behavior
Coronavirus COVID-19 USA Peak Caseload Prediction with daily tracking updates
- Evaluate progress toward curve flattening
- Independent assessment of peak case time and magnitude
- Evaluate alternatives for removing lockdown in minimum time
Each day I will update the known cases, and if warranted update the transmission rate (which would reflect lockdown effectiveness). The model will update to show revised date and magnitude of peak caseload, and risk of encountering an infected person if venturing out.
Highlights for 3/27 are
- Ability in the model to simulate removal of lockdown on a particular day (reverting to herd immunity as a percentage of original transmission rate) and several cases shown in plots. Optimal removal appears to require a week on / week off strategy.
- Flattening is working too well. Not enough people are getting sick to develop immunity. Lockdown removal any time this year (except by the week on/off strategy) will renew the catastrophe, but continuing lockdown will ruin the economy and people's future lives.
A near-optimal strategy (based on prediction with actual data through 3/27) is to release lockdown for one week on June 8, followed by one more week of lockdown, then permanent unlock. If you think you can do better, download the spreadsheet and go for it!Download Spreadsheet
- Please credit the source if using this analysis tool.
This model now automatically adjusts transmission rate based on previous days new cases and herd immunity (percentage of population that already had it). It does not predict policy variations, but assumes continuation of current day's policy. With variation in future lockdown, there should be follow on peaks which will not be predicted here. If flattening is partitioned (by state and city, for example) and more successful in some places than others, you would get a separate series of ever lower peaks like below, which do not indicate new outbreaks, but later peaking partitions:
One can get doublel humps out of the model by reverting to the initial transmission rate less herd immunity to simulate removal of lockdowns and social distancing. I will lmake some revisions and plot a few cases tomorrow.