I was looking over the data, specifically the slow ramp down in US data. What you will notice is that it is bumpy - up/down/up/down. What you might not notice is that the period is 7 days, and further, that the number of cases always hits a low on a Wednesday/Thursday, then grows until a Sunday/Monday, and then falls again until the next Wednesday/Thursday. (Interestingly, between yesterday and today, the worldometers switched to a 7 day average from a 3 day average overlay of the daily data which to a certain extent makes the periodicity a little less obvious in the continuous overlay)
.
US_daily_active_cases_05062020.jpg
Likewise, this same periodicity can be seen in the graph of daily deaths. Only it is delayed approximately 11 to 12 days.
US_daily_deaths_05062020.jpg
the curve tends to grow for 4 days and decay for 3. One possibility is that this periodicity is driven by the work week interaction of essential workers, people out shopping etc. Average incubation is 5.2 days, but the delays from first symptom to test to test result can be another 3 to 7 days. So it it very hard to say for sure what sort of weekly interaction is showing up here, but a fixed period of 7 days almost certainly points to some sort of correlation with the work week itself, which would tend to imply periods where we are out and about (work days?) produce more spread than periods where we tend to be at home with just our local family (weekends?)
I tend to think we are seeing the effect of social distancing on a weekly basis where periods of high interaction clearly show up as weekly peaks in the number of new cases per day, but there are a lot of variable that I don't have statistics for (when do people shop the most, what % are essential workers going to work 5 days a week vs 7 days a week (e.g. hospital workers) etc. What are the average delays between symptoms and testing/testing and test result. I can find ballpark guesses (like test result take from 2 days to a week) but that doesn't really help much.
What does seem to line up with the statistics is the delay between active case peak and death peak of 11 days or so. worldometers lists 14 days as average from first symptom to death giving an average 2 to 3 days delay for diagnosis -which seems in the right ball park.
If this periodicity represents the result of weekly increases in interaction under shelter in place just due to essential shopping and essential workers interacting, it does not bode well for relaxing those restrictions. That will increase interactions, and as it is we are barely on a ramp down. Increase those interactions by a factor of even just 2 by opening up and that downward slope will shift to an upward one.
I welcome any comments or alternate explanations for the 7 day cycle and active/death peaks.
.
US_daily_active_cases_05062020.jpg
Likewise, this same periodicity can be seen in the graph of daily deaths. Only it is delayed approximately 11 to 12 days.
US_daily_deaths_05062020.jpg
the curve tends to grow for 4 days and decay for 3. One possibility is that this periodicity is driven by the work week interaction of essential workers, people out shopping etc. Average incubation is 5.2 days, but the delays from first symptom to test to test result can be another 3 to 7 days. So it it very hard to say for sure what sort of weekly interaction is showing up here, but a fixed period of 7 days almost certainly points to some sort of correlation with the work week itself, which would tend to imply periods where we are out and about (work days?) produce more spread than periods where we tend to be at home with just our local family (weekends?)
I tend to think we are seeing the effect of social distancing on a weekly basis where periods of high interaction clearly show up as weekly peaks in the number of new cases per day, but there are a lot of variable that I don't have statistics for (when do people shop the most, what % are essential workers going to work 5 days a week vs 7 days a week (e.g. hospital workers) etc. What are the average delays between symptoms and testing/testing and test result. I can find ballpark guesses (like test result take from 2 days to a week) but that doesn't really help much.
What does seem to line up with the statistics is the delay between active case peak and death peak of 11 days or so. worldometers lists 14 days as average from first symptom to death giving an average 2 to 3 days delay for diagnosis -which seems in the right ball park.
If this periodicity represents the result of weekly increases in interaction under shelter in place just due to essential shopping and essential workers interacting, it does not bode well for relaxing those restrictions. That will increase interactions, and as it is we are barely on a ramp down. Increase those interactions by a factor of even just 2 by opening up and that downward slope will shift to an upward one.
I welcome any comments or alternate explanations for the 7 day cycle and active/death peaks.
Comment