Source: just a data based analysis by a friend Raju Chithambaram, Chief Architect in Intuit
THIS PANDEMIC WILL END, THE DATA TELLS US SO
We are seeing bigger numbers and unfavorable comparisons. 300,000 infections, 8000 fatalities, all serious numbers that should have us concerned. This post is an attempt to separate the facts from the fears, and bring calm through data.
First, let me get the noise out of the system. The US is not doing worse than Italy. The US population is five and a half times that of Italy. Comparing the US to Italy would be like saying the UK, France, Germany, Spain and Poland combined did worse than Italy.
Secondly, we see vanity headlines like "Fatalities now exceed 9/11 fatalities". This is irrelevant to the mitigation plan, unless we plan to bomb the virus.
Full disclosure -- this is a data model view (useful for correlation and trends). This is *not* an epidemiologist view. And even the data models are constrained because the data has covariance due to non-uniform testing and reporting. But the sample is still significant enough for us to learn.
Now coming to the data, which even amongst this gloom, offers us hope. If we measure the US as a whole, the single most important measure of spread is the daily new cases as a percentage of current cases. This was at a catastrophic 30% as recently as March 19 (which means doubling every 2 1/2 days). Thanks to the many measures we have put in place, that number has come down to about 12% daily growth (doubling every 6 days or so). We are doing a few things right.
What we could do better is bring consistency to our methods across states. Washington state growth rate is now down to about 5% daily growth, San Francisco bay area is getting there. These were the states and cities that were among the first to enforce "Shelter at Home" and other containment measures. The enemy is homogeneous in its methods, albeit not a visible one until it strikes. We could learn how these cities withstood this attack and quickly apply those measures elsewhere. Instead we are letting each state do it's thing. NYC is still up there. Lack of isolation in NY is the likely cause of new clusters emerging in NJ, Connecticut, Illinois. Louisiana and Florida are likely seeing the effects of not enforcing these measures earlier. The end result is a series of rolling peaks spread over time, where each state pays the price, and reacts to the peak instead of flattening and avoiding the peak. And there will possibly be secondary peaks in some states due to travel and transmission across state borders. If this continues, we will be dragging this out through August.
The call to action here is be the leader for your own communities and networks, and role model and enforce the social distancing, isolation and containment. The story told by the aggregate data is not as gloomy as the vanity headlines. We got it down from 30% to 12% -- that probably saved 10s of thousands of lives. This is still deadly serious, but we can limit the damage if we follow the measures taken by Seattle and the San Francisco bay area. En masse for four weeks
PS: The shorter version of this story reads "The second derivative of the exponent graph is declining, but not uniformly across states". But if you are like me, you cut calculus in college to go watch Sachin Tendulkar, Tom Cruise and Aishwarya Rai
THIS PANDEMIC WILL END, THE DATA TELLS US SO
We are seeing bigger numbers and unfavorable comparisons. 300,000 infections, 8000 fatalities, all serious numbers that should have us concerned. This post is an attempt to separate the facts from the fears, and bring calm through data.
First, let me get the noise out of the system. The US is not doing worse than Italy. The US population is five and a half times that of Italy. Comparing the US to Italy would be like saying the UK, France, Germany, Spain and Poland combined did worse than Italy.
Secondly, we see vanity headlines like "Fatalities now exceed 9/11 fatalities". This is irrelevant to the mitigation plan, unless we plan to bomb the virus.
Full disclosure -- this is a data model view (useful for correlation and trends). This is *not* an epidemiologist view. And even the data models are constrained because the data has covariance due to non-uniform testing and reporting. But the sample is still significant enough for us to learn.
Now coming to the data, which even amongst this gloom, offers us hope. If we measure the US as a whole, the single most important measure of spread is the daily new cases as a percentage of current cases. This was at a catastrophic 30% as recently as March 19 (which means doubling every 2 1/2 days). Thanks to the many measures we have put in place, that number has come down to about 12% daily growth (doubling every 6 days or so). We are doing a few things right.
What we could do better is bring consistency to our methods across states. Washington state growth rate is now down to about 5% daily growth, San Francisco bay area is getting there. These were the states and cities that were among the first to enforce "Shelter at Home" and other containment measures. The enemy is homogeneous in its methods, albeit not a visible one until it strikes. We could learn how these cities withstood this attack and quickly apply those measures elsewhere. Instead we are letting each state do it's thing. NYC is still up there. Lack of isolation in NY is the likely cause of new clusters emerging in NJ, Connecticut, Illinois. Louisiana and Florida are likely seeing the effects of not enforcing these measures earlier. The end result is a series of rolling peaks spread over time, where each state pays the price, and reacts to the peak instead of flattening and avoiding the peak. And there will possibly be secondary peaks in some states due to travel and transmission across state borders. If this continues, we will be dragging this out through August.
The call to action here is be the leader for your own communities and networks, and role model and enforce the social distancing, isolation and containment. The story told by the aggregate data is not as gloomy as the vanity headlines. We got it down from 30% to 12% -- that probably saved 10s of thousands of lives. This is still deadly serious, but we can limit the damage if we follow the measures taken by Seattle and the San Francisco bay area. En masse for four weeks
PS: The shorter version of this story reads "The second derivative of the exponent graph is declining, but not uniformly across states". But if you are like me, you cut calculus in college to go watch Sachin Tendulkar, Tom Cruise and Aishwarya Rai
No comments:
Post a Comment