Inspiration

As part of the Obama Administration’s efforts to make our healthcare system more transparent, affordable, and accountable, the Centers for Medicare & Medicaid Services (CMS) has prepared a public data set, the Medicare Provider Utilization and Payment Data: Physician and Other Supplier Public Use File (Physician and Other Supplier PUF). As we were searching through the particular governmental data, we noticed that much of the information was disorganized and difficult to understand. Additionally, Medicare provider data has not been visually mapped, to our knowledge.

What it does

The project, Geographical Visualization of Medicare Providers, plots the locations of individual providers and organization providers onto a map of the United States and provides detailed information regarding Medicare services and Medicare payments. Users can filter this information by the state of their interest and discover what providers are within their proximity.

How we built it

Our team created this web-app using the statistical programming language, R, and its packages, shiny and leaflet, HTML, and CSS. Additionally, our members used much of their own blood, sweat, and tears to ensure that the quality of this visualization is clear and easy to understand.

Challenges we ran into

Often times, it was an arduous task to deal with data of over 9 million observations and publish them on a web-app because it takes a lot of time and effort to computationally process the data. To overcome this, we took a sample from the 9 million observations, a sample of 30 thousand, and continued forward. Although it may still take a bit of time to load, the web-app ended up being completely functional and a reliable resource for users to understand Medicare data.

Accomplishments that we're proud of

None of us killed each other, which is fantastic! Additionally, all of us survived this weekend trek. However, we are quite pleased with this fully functional prototype after testing and attempting to crash it multiple times.

What we learned

Several of us did not think it was possible to come out with this web-app within the given time frame. Additionally, some of us are not as well-versed in R as others, so it was daunting to approach the task at hand. However, all of us came out with a new appreciation for R and a more developed understanding of how the language functions as a powerful statistical tool. Additionally, one of our members found out recently that R can produce efficient web-apps.

What's next for Geographical Visualization of Medicare Providers

The dataset that we dealt with had much more categorical variables that we could include to further filter the data such as "provider.type", "city.of.the.provider", and "HCPCS.code". Additionally, we aim to modulate the points on the map, by color and size, to better display differences between provider information so that users can effectively evaluate what provider is best for them. Whether it is an insurance company or someone receiving Medicare, we hope our future work on this visualization is beneficial and impactful.

Built With

  • r
  • shiny
Share this project:

Updates