Thread by @39Magilla

(1) Using Cell Phone Mobility data to estimate #COVID19 Transmission Rate (Rt) for the State of New York:

(2) State of New York Rt data from rt.live

rt.live/share/us/NY?r=…

(3) State of New York Mobility data from covid19.healthdata.org/united-states-…

(4) Estimating Rt using M:

IHME site gives "Change in mobility": C = (M - 100)

Rt(rt.live) =
0.6396 + 0.005154 x (100 + C(9.7 days ago))

Rt is a lagging indicator; M changes Exposures immediately, but won't be seen by rt.live for 9.7 days.

(5) More about lag in Rt from rt.live

The actual Rt will change immediately w/the change in Mobility, but rt.live estimates Rt from test results that contain an est. 9.7 day lag.

The lag is due Incubation, Symptom Progression & Test turnaround.

(6) Furthermore, Confirmed Cases tended to be tied to Hospital triage during the New York City breakout, but as testing expands, this will no longer be the case.

The relationship between Confirmed Cases and Mobility will change, but this correlation may still hold.

(7) In summary, the Mobility estimate calculated by @IHME_UW from Google, Facebook, Descartes Labs and SafeGraph data, may provide a good secondary signal to use for inferential control of the #COVID19 Transmission Rate (Rt).

(8) PS: To calculate a 9.7 day lag you need to use a weighted average of values. A simple solution is to use 30% of the 9-day old value plus 70% 10-day old value.

As an improved solution, the weighted average can also provide filtering.

(9) PPS: For New York, a Rate of Transmission (Rt) of 1.0 Corresponds to a Mobility index of 68-70%, or Change in Mobility of -30 to -32%.

In order to prevent exponential growth of the virus, New York should keep its Mobility index a safe margin below this, say -35%.

(10) Currently, the Change in Mobility is about -65%. You can keep an eye on this data at:

covid19.healthdata.org/united-states-…