Merge HIFLD data into HCRIS data

[lossyrob]

See issue comment https://github.com/daveluo/covid19-healthsystemcapacity/issues/14#issuecomment-599885877

This data should be merged into usa_hospital_beds_hcris2018_cleaned3.geojson in the data folder. The 'BEDS' field should be renamed to 'Total Licensed Beds'. Unclear what other columns should be carried over - perhaps an object inside the feature properties that just has everything, with a property name hifld?

Since there are lat lng coordinates in the HIFLD data, we can do a spatial join on buffered points to match the datasets.

[daveluo]

I would carry over all of the nonredundant columns from HIFLD. This dataset seems to be purpose-made to facilitate emergency/crisis management, i.e. with columns like HELIPAD, STATUS, TRAUMA, TYPE, OWNER

For example, STATUS == CLOSED facilities we can use to flag facilities that can be re-opened to add capacity.

[aaronxsu]

From https://github.com/daveluo/covid19-healthsystemcapacity/issues/14#issuecomment-599571862, these are the target statistics we're aiming to derive with this data:

• total licensed non-ICU beds & % of these staffed
• total licensed ICU beds & % of these staffed
• total number of vents & % of these available

[daveluo]

Note that we're mainly been focused on acute care capacity/bed supply for adults (these hospital types are often labeled as "Acute Care Hospital" or "Critical Access") right now but there are many other facility types in these datasets (like Peds and Womens' hospitals, nursing facilities, long term rehab). I'm guessing the processing and data merge is mostly the same amoubt of work whether we keep all facility types in or not? If so, preference is to keep every facility in the combined dataset and just not expose it to view for now

[daveluo]

Heads-up re: the HFILD data and some potentially missing data points (which we can quickly crosscheck against our HCRIS data summed by state or county):

thanks to https://forums.fast.ai/t/help-with-algorithm-for-covid19-relative-risk/65328/15:

About HIFLD, we are only showing in NY 1/10 of the beds they actually have, so there may be bad data, or I may have done something in preprocessing. I’m going to take a closer look soon but just make sure your data for NY State is checked/working.

[daveluo]

The type of facility is determined from the last four digits of its Medicare provider number:

Facility Type	last 4 digits of PROVIDER_NUMBER in HCRIS data
Short Term Acute Care	0001-0899
Childrens	3300-3399
Critical Access	1300-1399
Long Term	2000-2299
Psychiatric	4000-4499
Rehabilitation	3025-3099
Other	none of above

See HOSP2010_README.txt for more reference documentation about our HCRIS reported data (which is for all Hospital facilities as defined here and who submit that "2010" form: https://www.cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/Cost-Reports/Hospital-2010-form). From "HOSPITAL2010-DOCUMENATION.ZIP" in the above link