Skip to main content

At Veramed, we’re passionate about data visualisation because we know it’s one of the best ways to communicate data. In this article, Huw Wilson, one of our junior Programmers, describes how he developed a data visualisation based on a real-life case study. The data in this article is fictional but has the same structure as the original data. Learn how leveraging data visualisation principles can help you to improve your graphs!

The data was originally presented in a coloured table like the one below. It showed the results of a test for patients in a two arm study with 2 stages over 20 visits. If the result of the test was negative, only ‘neg’ was given, but if the result was positive, the value was given. In the 2nd stage of the study, all patients took the active treatment and for a few patients, the route of administration changed from intravenous (IV) to subcutaneous (SC).

The table is clearly very difficult to understand. Not all of the information can fit onto one page, and it’s difficult to follow an individual patient across the study. Most noticeably are the number of colours. Colour is used to indicate several variables, and consequently the ink-to-data ratio is very high. But how can we reduce the number of colours whilst keeping the same information?

In pattern recognition theory, colour is one way of suggesting an association. Another is through the use of proximity. By separating the treatment arms, we can imply the difference between them without needing colour. The graph below does this and instead uses colour to highlight the actual information of interest in the table – the results.

 This visualisation doesn’t yet show all of the information contained within the table, but we can already see how much easier the data is to understand. We can also be more strategic with our use of colour by highlighting the positive results with magenta and pushing the N/A results to the background with grey. 

Dose status is missing from the visualisation, and this is what we will focus on next. Since dose status can change at each visit, we can indicate this through the appearance of the lines. Dotted or dashed lines could work here, but I think changing the thickness of the lines creates a more striking effect.

The original table only showed dose status when a patient was not dosed and not assessed. In comparison, the above graph indicates dose status at every visit, but the focus of the visualisation is still on the test results. Including the values from the positive results also draws further attention to them. 

The only information now missing from the visualisation is the study stage and the route of administration. Since all patients enter the 2nd stage at the same time, we can indicate this by enclosing the area with a rectangle. As the route of administration only changes for a few patients (and doesn’t change again), a marker can be used to indicate the point of change

Not only is our final visualisation sharper and cleaner than the table, but it also contains all the original information and more. We can also start to identify features and trends in the data.  For example, why do some patients have repeating positive results, and why are there fewer positive results in the 2nd stage? 

In general, tables require much less thought to produce than visualisations. But often they are ineffective. With data visualisations, we have much more flexibility to make the data easier to understand, and most importantly, we can highlight the features that we are most interested in.

If you’re interested in learning more about data visualisation, or how we can help you to leverage your visualisations to greater effect, get in contact with us at Veramed!


Huw Wilson