Thread by @39Magilla

(1) Initial State Ro Values from @youyanggu model correlate moderately with State Population Densities.

(2) One of the most important properties for any infectious disease is the basic reproduction number, known as Ro. Rather than pre-setting this value based on assumptions, the @youyanggu model is able to learn the value that most closely matches the data.

(3) For the United States, New York had the highest initial Ro value; New Jersey was second. Other states had much lower initial Ro values including Arkansas, Mississippi and Montana.

(4) In an earlier thread, I theorized why Initial Ro values might vary, including the idea of "seed cases", "urban/suburban/rural" classifications, and "population densities" within classifications.

(5) I decided to see if one hypotheses could explain some of the differences seen between the Initial Ro values, so I plotted the values against each states population density.

(6) After omitting Alaska, because its overall population density was not indicative of its inhabited region, and adjusting New York by using NYC's population density in place of the state's, I found a moderate correlation.

Beware: Semi-log paper.

(7) The correlation showed the Initial Ro value increased by a factor of 1.3 X (10 to the 0.1155) for each 10-fold increase in population density.

1 => 1.0
10 => 1.3
100 => 1.69
1,000 => 2.20
10,000 => 2.86
100,000 => 3.71

(8) This isn't a proof, but it does suggest that people living in low population density locations will be able to control the rate of transmission with fewer restrictions, because the initial Ro value for their locale is naturally lower.

(9) Credits:

@uscensusbureau
@youyanggu

(10) PS: State Population Densities are not very indicative of specific outbreak conditions. For example New York State's density doesn't reflect the conditions in New York City well.

For this correlation this is just an additional source of variance.

(11) I just liked this diagram; it's ANOVA statistics art.