Here’s some reasonably good news, the rate of spread of COVID-19 is decreasing significantly in King and Snohomish County, Washington. Here’s a plot that illustrates the reduction in spread based on new cases.
The yellow, circular data points are a count of the actual cases as reported by King and Snohomish County Department of Health. The blue, red and green lines with constant slopes represent the exponential increase in COVID-19 infections predicted using an exponential model with doubling times that were altered to estimate the effectiveness of social distancing.
The blue (baseline) slope predicts the number of infections based on a doubling rate of 6.2 days; this is the case if no social distancing methods were employed. The actual cases (yellow) closely tracked the blue slope from 11 to 17 March, 2020. In the next week interval (18-24 March, 2020) the slope of the actual cases deviated from the blue line towards the red line which was an estimate of social distancing becoming more effective; that is manifest as a 25% increase in the doubling time to spread the virus.
In the last few days (28-30 March, 2020), the number of actual cases appears to be flattening out again. If that trend continues, that’s very good new for King and Snohomish County as it indicates significant reduction in the spread of the COVID-19 virus.
Update 31 March 2020
Here’s a different way to look at the data, suggested to me by my colleague Luis. This visualization based on a a trajectory method developed by Aatish Bhatia and Henry Reich using average weekly data plotted against cumulative data. [3] This is a lagging indicator, so I expect it to be delayed in signaling a downturn. That said, it’s likely to be a strong, positive signal.
Notes and References
[1] I derived my simple model as described in my post https://brewbooks.blog/2020/03/25/social-distancing-is-reducing-covid-19-burden-in-king-and-snohomish-counties/ using results originally developed by a team from the Institute for Disease Modeling with some partners: Klein, D., Hagedorn, B., Cliff Kerr, H. H., Bedford, T., & Famulare, M. (2020). Working paper – model-based estimates of COVID19 burden in King and Snohomish counties through April 7, 2020. Institute for Disease Modeling.
[2] Here’s a link to the underlying data and model I am using. Updated Covid19 Data SnoKing J Brew I will be updating this daily until 7 April 2020
[3] I encourage listening to Henry Reich’s Minutephysic’s YouTube video
and Aatish Bhatia’s website https://aatishb.com/covidtrends/
brewbooks 
I wonder if decreasing the X to Y axis ratio (maybe closer to 1:1) would make the change in slope more obvious.
Also be kind of fun to follow the first and or second derivative trends out to see when we might start coming down the other side.
LikeLiked by 1 person
Hi Luke
Good suggestions, I played a bit with the scaling and the trade off is that I want to continue this graph until 7 April (that will be 28 days in my simple exponential model).
I have looked at the first derivative quite a bit, see https://brewbooks.blog/2020/03/27/measuring-social-distancing-effects-in-reducing-covid-19-spread/ The second derivative is something I need to look at.
LikeLike