Data Cultivation

[bayoumi17m]

Find more data sources that showcase the following:

1. Number of COVID+ patients that are inpatient with sever status (hospitalized)
2. Number of COVID+ patients that are in ICUs
3. Number of daily hospital admissions of COVID+ patients
4. Number of COVID+ patients on ventilators

Ideally, we can stratify by age or other basic demographics (location within a region).

[bayoumi17m]

There is a lot of data online already for the US + for Other countries if we are willing to search through foreign language sites and use some google translate. Municipal websites (usually have github links) have been a good source. Another check, surprisingly is news articles!

From Iran: https://en.radiofarda.com/a/exclusive-67-000-suspected-coronavirus-cases-hospitalized-in-iran-/30515682.html

news reports can help make sense of more 'official' data if, e.g., columns are not well labeled.. did they mean 'hospitalizations' as census or as cumulative?

For the Italian data they have, they meant cumulative - they found out by talking with the owner of their data. They believe that the Spanish data is the same. In the Wuhan/Hubei data, they mean census. It is often hard to tell. But comparing to the numbers on given days in news stories can help make sense, and also can serve as datapoints from regions that don't post official numbers).

[bayoumi17m]

Data they currently have:

•   Italy data, which their health ministry posts in this github account: https://github.com/pcm-dpc/COVID-19
• Wuhan and Hubei and other China data, from Li et al: https://dash.harvard.edu/handle/1/42599304 and https://github.com/c2-d2/COVID-19-wuhan-guangzhou-data, which a Chinese colleague has collected more of from the Wuhan municipality site: http://wjw.wuhan.gov.cn/
•  Limited South Korea data, which was shown in a Wall Street Journal article: https://www.wsj.com/articles/how-south-korea-solved-its-acute-hospital-bed-shortage-11584874801
• Data for >36 US states (so far) including NY:  https://covidtracking.com/data/
• Spain data from here: https://covid19.isciii.es/

[bayoumi17m]

Note, even having 1 or 2 datapoints from a country ('e.g., there were 6k patients in the hospital on March 1, and 12K on April 1'), and knowing when they shut down or did other measures, is probably enough to make the data useful. That's why putting some work into finding news stories can be useful.

[bayoumi17m]

They have been very interested in comparing timing of different interventions across countries to the slowing or stopping of exponential growth of case counts according to the John's Hopkins COVID-19 data, which are here:

https://github.com/CSSEGISandData/COVID-19

I believe that the insights into what slows / stops the curve and timing (how long does it take?) gainable from that analysis could help inform the other task as well.