Text Severity

Tender description

We are inviting quotations to develop a text mining tool which allows us to automatically identify those comments where the experience captured is of such severity that they should be immediately flagged for review or further action. We will provide a data set of approximately 30,000 - 40,000 patient experience comments captured from 2014-2016 from our Share Your Experience website. These are coded for severity with about 300 flagged as serious. Similarly we have 22,000 comments from NHS choices for 2014-15 of which about 100 are flagged. The severe comments are associated with organizations (mostly acute hospitals) and also coded for sentiment. We continued to collect this data after 2016 but have not coded it but can make it available. Although this tool is to be developed on available data sources we aim that it should be capable of being extended to other feeds in future which may include social media sources. We are, therefore, seeking a single organisational partner with considerable knowledge of: • using high volumes of information and, in particular, qualitative intelligence from people who use services and the public about the quality of their care. The provider should, therefore, also have knowledge of the range information sources where people comment on the quality of their care (including specific sites such as NHS Choices and Care Opinion and more general locations such as facebook or twitter). • analytical techniques to identify severity in high volumes of qualitative intelligence. • significant risks to good quality care and how they can be identified. This should take into account both the sentiment of any comment and, importantly, the risks associated with the events being described.