Thursday, 9 August 2012

The making of Infographic #2

I thought readers may be interested to know how my second infographic evolved, particularly as it changed quite considerably from my initial conception.

A few months back I attended a course by Andy Kirk on data visualisation, in which he highlighted the importance of conceiving your visualisation design, rather than just jumping head first into it. Having already familiarised myself with the data back in Infographic #1, the purpose of the second one was to really highlight the sheer variety of side effects that these five drugs alone could potentially cause. In my mind I already knew that I did not want to use any numbers, it would be kind of meaningless. I also knew that I wanted to show the circle of side-effects caused by one drug and the treatment of that side-effect by another. For example, Omeprazole is used in the "treatment and prevention of...NSAID-related ulcers"*. An NSAID is a non-steroidal anti-inflammatory drug, an example of which is Aspirin. Interestingly, BNF does not list ulcers as a side-effect per se (hence why it is not in the diagram) but chooses to use the less specific description of "gastro-intestinal irritation or gastro-intestinal haemorrhage"; more of a catch-all I suppose. So I had a few conceptions in my mind (and crucially had already ruled out some ideas), but visualisations need to sometimes be seen on screen to get a feel for whether it could work or not.

My first idea was as a table, with red pills showing side-effects, green pills showing conditions that are treated/prevented. The five drugs were the columns, the conditions the rows. I quickly realised that there were too many conditions for it to be readable. Perhaps with <20 conditions, this could have been do-able.

My next thought came to me in the middle of the night (they normally do), and I was thinking of venn diagrams but using body organs instead of circles, with conditions in one circle for treated by, then conditions in another that were side-effects, with the overlap being those conditions that are in both categories. On drawing it out on paper though I quickly discarded that idea: I couldn't figure out how to distinguish the five drugs.

Next up was a mindmap idea. This was the most time-consuming as I didn't have any network diagram software which would automatically pick up the connector line with the text/circle (as it does in Microsoft Word when you're creating a mindmap). So I fiddled around with this for ages, having created vector images of body organs (again something that took a LOT of time!!!) and then reluctantly gave up on this, as there were just too many lines. I really should have figured this out from the table I did!!! I knew the text would never be big enough to read, and it didn't have that "wow" factor yet.

My penultimate idea was to have the five drugs at the centre of the infographic each discerned by an oval (so I could make the middle look like a pill just to over-theme the whole thing!). This really started to look promising as I could represent side-effect as a red dot and treated by as a green dot, and it would allow multiple dots for one condition. The one problem was the alignment of the lines leading from the conditions, it was still too jumbled.

So it was then I thought of switching to a circle instead of an oval in order to overcome this, which is what you see in the final version. I then grouped conditions by the body system (seemed most logical to me!). Classifying the conditions itself was a little challenging given that I am not a medical practitioner and that Google can only take you so far, so some of them were my best estimate!

So crucially what lessons have I learned?

  • First and foremost, I've got to keep it 'rough and ready' until pretty much the final draft. What I mean is, I spent ages early on creating these nice vector images of the body organs but they ended up not even being a critical part of the infographic. The main thing should have been to get the outline down on Illustrator, and then the elaboration should come last. What was important was realising how many conditions I had to deal with, and then ruling out visualisations quickly based on that. 
  • Second, never underestimate the onerous task of data collection if the data is unstructured. The data here was from a website, so I had to get it all in an Excel sheet, with one condition per row. I faffed about it with it for ages, looking up each condition on Google. I did this for all of the top ten drugs, when I only use the top five. A bit of wasted time, but not disastrous. 
  • Third, I've learnt a lot more Adobe Illustrator tricks which I will use in the future. The best one for this was using a pie chart of 88 segments (how many conditions are listed on the infographic) to help align all of the lines. Using the alt-drag to copy shapes, and the eyedropper, and the various align and pathfinder tools certainly helped to speed things up a bit also. 

Now on to the next project.........

*From the BNF description of Omeprazole:

No comments:

Post a Comment