The idea start with me thinking 'let's throw all hospitals in the US on a map to see what happens.' Turns out there are huge swaths of the country where the nearest hospital is up to 104 miles away. The data on hospitals came from CMS, and I converted the addresses to latitude and longitude coordinates using a Python library. From there, I imported that to ArcGIS, put 30 miles buffers around each hospital and dissolved and clipped the map. I then calculated the area of the area on the buffered tile and exported the map for the story. My colleague Ciara Frisbie helped me interview sources living in these areas.
Partner Andrew Briz and I made a Twitter bot that if someone tags in a reply to a link on Twitter will take the words in the headline, run them through PolitiFact and see if anything that's a close enough match comes up. If so, it will reply to the Tweeter with the link. This is a project we hope to expand further with Snopes and eventually machine learning.