A grad school class in epidemiology and cartography. Pandemics are fundamentally about population density, and therefore geospatial in nature. During grad school, I developed a toolkit of epidemiology mapping tools that could connect to EHRs. I had been using the toolkit to analyze some Social Determinants of Health data and Opioid Epidemic data, and have shopped it around the past couple of years, with little success. Covid19 has suddenly made my toolkit extremely relevant.
Please check out the online demo. We've just launched, and there will continue to be updates pushed on a regular basis for as long as this project is needed.
What it does
Queries hospitals for COVID19 related data based on LOINC and SNOMED codes, does a patient demographic lookup on the results, geocodes the results, generates a geojson file, and displays it on a heatmap in Google Earth.
How I built it
I actually wrote an earlier version of the mapping software for a grad school class. The original version mapped various datasets found from the City of Chicago data portal, such as the EPA Toxic Inventory, location of grocery stores and restaurant violations, crime incidents, etc.
The major challenge in this hackathon has been a) modeling the progression of COVID19 using Synthea, a synthetic patient generation tool, so we could have a sample dataset to work with b) identifying the specific LOINC and SNOMED codes that the hospital EHRs will be using, c) and updating the map to be COVID19 specific.
Challenges I ran into
Coffee. I've been working on this pretty much round-the-clock for three weeks now.
Accomplishments that I'm proud of
The Synthea Covid19 module.
As for the software, it's very stable and much faster than previous versions. The code is very clean, and we even have a quality control script for it.
I'm also quite pleased with the aesthetics. The maps grab attention and communicate important information.
What I learned
Clarity of a simple one-page app. I've had a bit of second system syndrome with a prior PHR project. This has been real nice to have the focus of a clearly defined problem and knowing clear sequence of steps to create a solution for it (even if that sequence of steps is a hundred items long). Learning some good lessons in terms of product design and keeping things simple and maybe product-market fit.
What's next for COVID19 on FHIR
Registering it with the Epic App Orchard, Cerner Code Gallery, and Allscripts App Store.
Pilot Sites: Identifying hospitals (preferably in Chicago) to pilot with.
Adding custom workflows. The underlying mapping technology has applications for just about every stakeholder, but there seems to be particular need at the population health level (placement of testing centers, tracking overall spread), the field dispatch level (EMTs, police, social workers), and the individual patient levels.
Querying EHRs for bed and ventilator capacity. We have the Device, Location, Measure, and MeasureReport resources and UI components available for developing an automated bed capacity utility.
External Data Inputs: Adding a REST microservice so we can receive data sent to us. A data management page, so people can drop spreadsheets into the utility.
Business Meetings: We have conversations with Allscripts and the State of Illinois scheduled for this week.
Business Development: Determining a pricing point. Determining how to manage a Docker infrastructure that we use that runs on AWS, but isn't necessarily interested in getting involved with HIPAA.