We were inspired by the simple question we have heard from neighbors, patients, citizens, physicians, and public health leaders: "What is being done in my area and how will it affect how COVID-19 will spread here?"
What it does
Our COVID-19 policy map is constructed by crowdsourced data about policies currently in place around the country, and indicates visually at a county level what policies are currently enacted.
How we built it
We are building a custom frontend for data collection on policies from the general public in Angular.js,with designs constructed in Figma. We are designing a data collection tool to be simple and accessible to the general public to input data from their ZIP code. We collected crowd-sourced data using this Google Form, while we are in the process of building our custom frontend: linkWe wrote a Python script to process the text data from the Google Form and encode it categorically for use in data science applications. This data can also be exported as a CSV or JSON. Finally, we used the Google Maps API with a GeoJSON library for county IDs to create the framework for a map-based visualization of the policy data.
Challenges we ran into
We initially sought to build a predictive tool for the spread of COVID-19, and found that the major confound that limited our ability to model the spread was the lack of structured data on policies currently in place. Thus our idea for a policy map was born of this challenge. Another challenge was visualizing the data at a resolution that is local enough to be relevant to an individual citizen without demanding an unrealistic amount of crowdsourced data. We settled on county level as a happy medium between ZIP codes and states, particularly since many policies are implemented at the county level. Finally, we wrestled with the ethical challenges of sharing information that can be easily misinterpreted with the public in an informative context. We decided not to immediately pursue building a predictive _ risk map _ by county because without the policy data, the chances for misinterpretation and misrepresentation are significant.
Accomplishments that we're proud of
What we learned
We learned that the user is of foremost importance when building a product, particularly with health related information. Designing a map for the public to use to inform themselves ended up being significantly different in terms of features from designing a tool for epidemiologists, data scientists, or policy makers - all niche users we considered.
What's next for Hikma Health COVID-19 US County Policy Map
We plan to crowdsource enough data to populate our map (at least 1k respondents from 200 counties). We will superimpose our map on the COVID-19 epidemiological map to allow for visualization of the effects of particular policies on cases and fatalities. Finally, using our structured policy dataset, we will be able to incorporate a predictive model for the cases of COVID-19 expected in a particular county that incorporates vital data about the policies in place.
Try It out
angular.js, google-maps, pandas, python