In healthcare we’re data rich, but information poor. There are lots of surveys, studies, and healthcare data available to us through different programs, but very rarely do we see it presented in way that is clear and understandable to an audience that isn’t familiar with the subject matter. I’m going to highlight three of my favourite visualizations today – be sure to share your favourites in the comments below or on Twitter!
The drug overdose epidemic is one that has been growing rapidly over the past few years, with wide geographic variation. Rates at the state level vary dramatically, with ~ 6.9 deaths per 100,000 population in Nebraska (the lowest in the country). Meanwhile, rates in West Virginia were almost 6 times higher, with 41.5 deaths per 100,000 in 2015; a shocking 17% increase from the previous year (click the image below to see where your state ranks).
Last week the NY Times presented the drug overdose epidemic in a way that was incredible insightful and profoundly effective. In a piece titled “You Draw It: Just How Bad Is the Drug Overdose Epidemic?“, you were given historic data and were asked to plot forward how many Americans have died every year from car accidents, gun deaths, and other major causes of death. From this, you can project forward and see how your estimates compare to what is actually being observed.
What I really liked about this was how the website was set up to engage with health data in a way that is very different from static graphs and visualizations, and forced you to think about what you think it happening, then presented you with data that either corroborated or went against what you expected. For example, I was surprised to see how high deaths from car accidents were, but that could be for a multitude of factors.
Another really good example of visualizing data is the Census 2016 Demographic DJ by the CBC. This was a really innovative way to see how a reader fits into Canada’s demographic landscape, in terms of how many people are older than you, how the number of people in your age group changed since 2011, and other demographic details. As more information is released, they will continue to build out other visualizations that convey this information even more effectively.
Now, healthcare and public health are the obvious focus of this blog, but one visualization that I absolutely love that uses visuals incredibly effectively to lead you through a story is this one on The Pudding: Are Pop Lyrics Getting More Repetitive?
This image takes a relatively common theory – that modern songs are more repetitive than older songs, and tests it with data. However, the way they do this is by using an algorithm called the Lempel-Ziv algorithm. The piece opens with a description of how the algorithm works – showing you as you scroll through how duplicated words are eliminated. It highlights in a very easy to understand fashion how a very complicated algorithm works, and the effect of it. For example, it shows how you could compress the entirety of Around the World by Daft Punk into 61 characters from 2,610. But it then breaks things down even further, looking at trends over time, how this is driven by certain artists, and finally how much variation there is by individual artists depending on the song. I highly recommend taking 3-5 minutes and going through it.
Public health, and medicine in general, can learn a lot from how other industries convey important and complicated information. Even when the data isn’t related to health information, we can still learn from it to make our findings engaging to an audience who may not be immersed in the subject matter, but would like to know more if it is presented in a compelling and engaging way. Other industries are doing this already; we just need to learn from them and tailor our content to a public health audience.