Patient Reported Experience Survey: Engineering of Natural Text v1

  • Research type

    Research Study

  • Full title

    Patient Reported Experience Survey Engineering of Natural Text (PRESENT): Developing practical automated analysis and dashboard representations of cancer survey freetext answers

  • IRAS ID

    180736

  • Contact name

    Carol Rivas

  • Contact email

    c.a.rivas@soton.ac.uk

  • Sponsor organisation

    University of Southampton

  • Duration of Study in the UK

    1 years, 9 months, 31 days

  • Research summary

    Surveys of patient experience can help NHS service providers see where and how to do better. Some survey questions require tick box answers (for example ’yes’/ ’no’), or ratings of services on a scale. These are quick and easy for providers to evaluate. Some questions, called ’freetext comments’, invite people to give more detail in their own words. Analysing these takes months and is labour intensive, which stops freetext being well used. We will develop and test a better, faster way of analysing survey freetext, called ‘text engineering’ with a ‘dashboard’. Our research has three parts:

    1. STAGE 1 TEXT ANALYSIS or ‘ENGINEERING’. We will adapt a special computer programme that rapidly analyses general text using the natural language rules humans use to talk and write. We will add healthcare words to its vocabulary and modify its rules so it can accurately process language the way people write it in health survey freetext comment boxes. Using a questionnaire, we will ask 150 ‘healthcare stakeholders’ (patients, partners/carers, NHS staff) to contribute words and also to tell us which aspects of healthcare delivery to prioritise, so our entire process will reflect their perspectives. We will use our modified programme to analyse freetext from a national cancer survey (CPES) and check accuracy against results from more traditional analyses we have already done.

    2. STAGE 2 DESIGN OF A ‘DASHBOARD’, an interactive website to visually summarise our new analysis alongside relevant data the NHS routinely collects. Around 100 stakeholders will help us design a dashboard for everyone, through structured large groups and interviews.

    3. STAGE 3 PROOF OF CONCEPT: We will test our approach with another cancer survey and ask 15 staff in 3 UK NHS Trusts to try out our dashboard and give feedback. We will make improvements as necessary.

    Our CPES work will be freely available.

  • REC name

    North of Scotland Research Ethics Committee 2

  • REC reference

    15/NS/0104

  • Date of REC Opinion

    22 Oct 2015

  • REC opinion

    Further Information Favourable Opinion