Safely making free-text notes available for health research

Health care data plays a pivotal role in transformative discoveries and enhancing patient outcomes. However, the journey to harness the full potential of health data is not without its challenges, such as those posed by clinical free-text records (including hospital discharge summaries and GP notes). These records hold valuable health-related information that sometimes cannot be found in coded - also known as structured - data. However, unlike coded data, free-text notes also contain sensitive information such as details of a patient that have the potential to make them identifiable. 

Natural language processing (frequently called NLP) is an approach that can be applied to extracts of these data to ensure they are made safe and secure before granting researchers' access. Exploring this area is Arlene Casey, who is a Dunhill Medical Trust Senior Research Fellow and the Principal NLP data scientist at DataLoch, which is a data service that provides access to extracts of data to approved researchers for projects in the public interest. Arlene's focus is on how free-text records could be processed to ensure people's identities are kept secure while enabling the data to be safely used for research. Integral to this development work is consulting with the public and being open and transparent in the methods used by DataLoch. Recently, DataLoch partnered with Ipsos Scotland in an initiative to undertake public consultation of privacy risk assessment in clinical free-text. 

In this podcast, Arlene and Ciaran Mulholland from Ipsos Scotland discuss a novel public consultation that has informed the decisions made by Arlene and her colleagues.

The Data Driven Innovation initiative is supporting hundreds of collaborations and organisations across Edinburgh and the south-east of Scotland. For more information follow their page on LinkedIn, find @datacapitaled on X (formerly Twitter) or go to the website ddi.ac.uk.