Option for adding Google Mobility data as an additive mitigation NPI

[santiago-afonso]

🙋 Feature Request

I'd like an option to quickly add Google Mobility data (https://www.google.com/covid19/mobility/) difference from baseline as a daily mitigation intervention.

🔦 Context

I'm trying to visually "fit" the model's predicted cases curve to the observed cases by fiddling with the assumptions of the epi parameters and the mitigation interventions. For a given set of epi parameters (mostly R0), making the predictions consistent with the observed case numbers (which are displayed by the tool) and other locally available data not displayed by the tool (such as ICU usage) is hard as it requires varying the mitigation intensity week by week. In regression analysis, mobility data captures most of the effect of the mitigation strategies such as the different lockdowns (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3581633).

Using the "mitigations" panel to "draw" manually what would quickly and more precisely captured by mobility data is not optimal.

😯 Describe the feature

A tick box in the mitigations panel would additively add Google Mobility data for the selected geographical area as an intervention. The subindices for workplaces and transit should be used rather than the other subindices that include things such as parks, given that we know that transmission is most likely to occur indoors and in sustained interactions.

Additively: the rest of the panel would work as today, but allowing for negative numbers. If the google mobility data for a given day is -80% and interventions for that day add to a -20% in the panel (which currently displays it's sign inverted for clarity purposes), the sum of the interventions would yield -80% -(-20%) = 60%. This would allow for further adjustments.

But the panel should mostly keep working as today.

Thanks!

[rneher]

adding mobility data could be fun. We could overlay this with the Rt graph. But I haven't looked into Google's API and how we'd match up keys for different regions.

[santiago-afonso]

The csv file is really simple

https://www.google.com/covid19/mobility/

CSV: https://www.gstatic.com/covid19/mobility/Global_Mobility_Report.csv?cachebust=6d352e35dcffafce
countries are ID'd by name and by 2-digit ISO country code. Country totals have missing data on other columns that otherwise indicate geographical region (state/provinces, county/municipality/department).

Here's the csv documentation:
https://www.google.com/covid19/mobility/data_documentation.html?hl=en

Thanks!

[TakumaKira]

Hi there!

I tried adding this feature. I added "ADD MOBILITY DATA" and "REMOVE MOBILITY DATA" buttons on the MITIGATION panel.
Demo
Changes

I think we need to discuss how we can calculate the transmission reduction from the mobility data. Currently, this simply takes an average of "Transit stations" and "Workplaces."