Cancer Patient Experience Survey: Using Computerised Text Analysis and Text Analytics to Explore Big Data

Abstract

This case study considers the way that sociologists and computer programmers, working together, developed an automated approach to thematic analysis (text analysis or text mining) of survey free-text comment boxes. We also developed a website powered by our text analysis but additionally producing text analytics to summarise findings. We used a cancer patient experience survey as our test case and an approach called rule-based information retrieval, rather than standard supervised machine learning. Linking automated text analysis of surveys and text analytics with a website toolkit in the way we did was novel and helps health care teams to consider the comments and make improvements to patient experience as a result. The way we asked health staff and patients to meet and talk together in specially designed participatory workshops to help us develop our toolkit was also novel for a machine learning study. However, we had to solve several issues to complete the study. Programmers and sociologists had to reconcile their very different ways of working. Ultimately the sociologists in the team developed a better understanding of what computational approaches could achieve, which was different from their original expectations, and the programmers learned about the added value to be gained from the workshops. We all learned about the importance of having a variety of study outputs, to reduce dependency for success on the novelty of the technology (given the dynamic progressive nature of the field) or on data access permissions, which inevitably take longer than anticipated.

locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles