Type 1 Diabetes: A Twitter Data Scrape


Post 6 by Lucy Allen

Twitter has been defined as a sort of ‘microblogging’ in allowing Its users to post short, 140 character posts on the ‘twitter-sphere’. The limited word count often means that witty and somewhat blunt opinions and statements are made and shared with followers and available to a worldwide community.

The appeal of Twitter is not only the freedom of opinion and expression but the ability to read content quickly and engage with users world-wide. An array of accounts manned by different companies, individuals and businesses means that live events are responded to and can be tracked in real time.

Screen Shot 2016-09-05 at 2.10.23 PM
(Buzzfeed, 2014)

Due to the array of users and audience members there are many different reasons and opinions expressed in ‘tweets’. The different identities and values of the different users are mapped out below:

new doc 6_1
Mapping the values and identity of different Twitter users (Lucy Allen, 2016)

The tone of voice used in a Tweet very much comes down to the user themselves and The type of user they are. For example, a business company’s tweet about it’s recent successes would most likely have a much more formal and matter-of-fact tone of voice compared to that of a young adult on the topic of a music festival. In saying this companies and even governments/politicians sometimes utilise a more informal and humorous tone of voice over social media platforms to attract a younger and more varied audience.

(Lindsey Graham, 2013)



With over 313 monthly million users and 500 millions tweets per-day, Twitter is a treasure trove of information and opinion. The development of data scraping tools such as the ‘Advanced Search’ Twitter function allows anyone to discover the data and information across the enormous platform. This allows people to examine and explore this data, pulling out interesting findings and insights.  I undertook my own data scraping task to investigate the presence of discussions about Type 1 Diabetes, excited see what I could deduce from my process and findings.


My process for data scraping utilised the Twitter Advanced search function as an add-on of Google Sheets. Whilst I began with this special function I ended up using the Advanced Search on the Twitter to enable more thorough data scrapes.

new doc 7_1.jpg
Process map (Lucy Allen, 2016)


My first search was very broad and included all the related words I had brainstorm in regards to Type 1 Diabetes. This broad search resulted in 170 pages of data and over 10, 000 results. From a brief glance at the results I did immediately recognised that there was a real mix of personal posts from people living with the disease, mixed in with news reports and research findings. I realised however that I needed to refine my data scrape if I was going to be able to properly analyse the data.

Screen Shot 2016-09-03 at 1.45.43 PM
Twitter Advanced Search 1

My next search involved searching fewer words but still keeping the terms broad by searching ‘Type One Diabetes’, ‘Type 1 Diabetes’ as well as those terms as hashtags. I got 450 results back, a much more realistic set of data however still to big to analyse and draw insight from.

Screen Shot 2016-09-03 at 1.57.42 PM
Twitter Advanced Search 2

Using the same search terms I wanted to narrow the date field of the data, limiting it to the past 30 days. I discovered that the Twitter Advanced Search add-on for Google sheets didn’t enable a date-based advances search. Instead I ran the advanced search on the Twitter website which did allow me to refine my search by date and time frame. The results weren’t as neatly collated as that of using the search in conjunction with Google Sheets however it did enable me to see entire posts, even those that included images and videos. I was able to spend time looking into the search results and made some insightful discoveries.

new doc 4_1
Search 3:  Tallying the types of Twitter posts on Type 1 Diabetes

I went through the collection of posts and started to tally what type of posts they were. I wasn’t surprised to see that most posts on the topic of Type 1 Diabetes were personal ones that saw people expressing day-to-day struggles of living with the disease. There were certainly fewer news and research based posts than I was expecting and far too many posts about Nick Jonas, a celebrity living with Type 1 Diabetes (I had to give him his own category…). There were of course a few ‘really?!’ tweets where people obviously had no idea about the disease as seen in the example below.

Screen Shot 2016-09-03 at 3.02.01 PM
Is Type One Diabetes Real… (Twitter, 2016)


Probably the most incredible discovery I made throughout my data scrape was the discovery of so many support networks and platforms for people living with Type 1 Diabetes to connect with. I was completely blown away and can’t wait to look into these further and discover the digital community that exists.

After I’d completed this area of investigation I was interested in looking at the results I got when searching Type 1 Diabetes imagery on Twitter, again with the Advanced search tool. After creating my own image archive last week I was really intrigued to see what types of images would come up. I found that most images were in regards to personal achievements of those living with Type 1 Diabetes or fundraising or campaigning. I noticed again and as discussed in my recent blog post ‘Type 1 Diabetes: Stakeholders and Visual Representation’ that there was a lack of images that can completely sum up what it is like living with Type 1 Diabetes.

Key findings

  1.  People are using social media to speak out about what It’s like living with Type 1 Diabetes
  2.  Social media platforms such as Twitter are acting as a digital support network for people living with diseases as well as other issues and concerns
  3.  There are still lots of people who are completely unaware of Type 1 Diabetes or lack any proper understanding of the disease
  4.  Most ‘tweets’ about Type 1 Diabetes came from Westernised countires
  5.  People living with Type 1 Diabetes are becoming increasingly frustrated with a lack of understanding and stereotypes present in society


In reflection I believe that there may have been more creative potential if I searched some less obvious terms such as ‘insulin pump’ or ‘Autoimmune disease’ etc as part of my data scrape. I also would have liked to have spent more time analysing and looking into the image archives of Twitter however this came as an afterthought to my initial process. I noticed when going back over my search results you can also generate positive or negative responses. It’d be interesting to gather and compare this data, looking at the different tone-of-voice these opposing posts take.

Screen Shot 2016-09-05 at 9.53.43 PM.png

I would also love to have the skill and time to create a Twitter Bot as Chris Gaul talked about in his lecture. It would be fascinating to interact with people posting on this topic and even pose questions to them or provide them with knowledge on Type 1 Diabetes. These bots have so much potential in engaging with the issue through a tweet and therefor engaging with users and people from all walks of life.

Like with all data there is so much potential in how it can be visualised and broken down. A fantastic pre-existing example of this is the #NoPricks campaign image which visualises the average amounts of injections a person living with Type 1 Diabetes has in a month.

#NOPRICKS Campaign Poster (Nopricks, 2016)

With the data and content I’ve collected It would be really great to visualise the breakdown of the types of tweets people post about Type 1 Diabetes as I think it says a lot about social media and sharing and support platform. It’d also be interesting to delve into this further by breaking down the types of personal posts and analysing the reasoning behind why someone may want to post this and the response/support each post gets. This then got me thinking about the possibility of mapping the online support network of sufferers living with Type 1 Diabetes to see the great extent of this network. An example of what this type of data visualisation could look like is seen below:

new doc 8_1.jpg
(Lucy Allen, 2016)


Gaul. C, 2016, ‘Twitter Bots’, Lecture, accessed 1st of September 2016, <https://online.uts.edu.au/bbcswebdav/pid-1384637-dt-content-rid-8098732_1/courses/87831/Week4Lecture_ChrisGaul_TwitterBots.pdf&gt;

Gil, P. 2016, ‘What Exactly is Twitter’, About Tech, accessed 1st of September 2016, <http://netforbeginners.about.com/od/internet101/f/What-Exactly-Is-Twitter.htm&gt;

Heugel. A, 2016, ‘Lindsey Graham Tweet’, 30 Hilarious Political Tweets, accessed 1st of September 2016, <http://twentytwowords.com/hilarious-political-tweets-that-will-make-the-internet-great-again/>

Muhammad M. U., Muhammad. I , Hassan. S, ‘Understanding Types of Users on Twitter’, Lahore University of Management Sciences, Lahore, Pakistan1 Qatar Computing Research Institute, accessed 1st of September 2016, <hsajjad@qf.org.qahttp://www.qcri.org.qa/app/media/4858>

NoPricks, 2016, ‘Campaign Image’, accessed 2nd of September 2016, <http://www.prweb.com/releases/2014/08/prweb12069199.htm&gt;

Sayce, D. 2016, ’10 Billion Tweets‘, David Sayce, viewed September 1st 2016, <http://www.dsayce.com/social-media/10-billions-tweets/&gt;

Twitter Advanced Search, Twitter, accessed 3rd of August, <https://twitter.com/search-advanced?lang=en&gt;

The Huffington Post, 2016, ‘Best Tweets’, accessed 4th of September 2016, <http://www.huffingtonpost.com/news/best-tweets/>

Tholepin. C, 2016, ‘Is Type One Diabetes Real’, Twitter, accessed 2nd of September 2016