Fig 1.
(a) View of interactive map displaying COVID-19 cases and outbreaks across schools in Canada, collected by COVID Schools Canada (covidschoolscanada.org). Map generated using MapBox (https://www.mapbox.com/). (b) Example view of code-generated tweet displaying summary statistics.
Table 1.
Quick tips to crowdsource data for public health advocacy.
Fig 2.
Workflow used by COVID Schools Canada (CSC) to crowdsource COVID-19 cases and outbreaks data across Canadian schools.
Data were collected using a combination of automated web scrapers (Tip 8), manual spreadsheet entry from a combination of school board and province-level websites (Tips 6 and 7), and from a Google form on the CSC website for individual reports (Tip 9). Custom R code was used to pull data from these varied sources, clean the data, and assign latitude/longitude coordinates to each school entry (Tip 10). The final CSV file was converted to GeoJSON format and pushed to GitHub. Mapbox was used to create the interactive map and automatically refreshed its view from the file on GitHub (Tip 13). Table 2 lists all major software packages used for building the pipeline. Map generated using MapBox (https://www.mapbox.com/).
Table 2.
Technologies used for Covid Schools Canada.
We recommend teams reach for tools they already know, and where there is an intractable gap, recruit technical expertise to help overcome it. See S1 Text for associated URLs.
Fig 3.
Example of centralized data storage for a crowdsourcing project and shared data entry table.
(a) Directory tree on Google Drive containing web-scraped data (“Schoolboard Daily Snapshot”), spreadsheets for manual data entry (“Nightly Data Entry”; see detail in b), and supporting documents for case reports (“Supporting Docs”). Created using Biorender.com. (b) Example of a manual data entry file for the COVID Schools Canada project. Fields are standardized across spreadsheets, and the “Instructions” tab contains a reference for data for each field. The sheet contains a master list of school names, addresses, and geographic coordinates, which team members copy-pasted to create new case entries and minimize errors due to typing (“Master List of School Codes”).
Fig 4.
Left: The call we put out on social media that led to the connection with Mapbox (Top Right). Bottom right: Javascript-based map created by partnership with Mapbox. Map generated using MapBox (https://www.mapbox.com/).