Overcoming language barriers in UK healthcare at HealTAC 2019

A panel of researchers, patients, clinicians and companies gathered at the Healthcare Text Analytics Conference (HealTAC) to discuss the state-of-the-art in processing healthcare free text when language barriers exist.

When a patient and the health professional who cares for them both have first languages other than English, what's the best way to conduct consultations, keep and share records? As this year’s UK Healthcare Text Analytics Conference was organised in the capital city of Wales, the conferences panel groups explored this issue in the case of the Welsh language.

The discussion, held in Park Plaza Hotel in Cardiff on 24–25 April, will support the development of new inspection methodologies, processes and reports to support the implementation of 'More than just words: Follow-on strategic framework’, which was published in 2012 by the Deputy Minister for Social Services in the Welsh Government.

The framework aimed to ensure that relevant organisations recognise that language is an intrinsic part of care and that people who need services in Welsh get offered them. This is called the ‘Active Offer’.

The intention of the follow-on strategic framework 2016 -2019 was to maintain momentum and build on the previous strategy, including practical support for implementing the ‘Active Offer’.

The panel explored technical means that can facilitate service provision in Welsh, including terminological resources, software localisation and machine translation, as well as a variety of application areas, such as prescriptions, mental health and allergies.

There were valuable contributions from local researchers, clinicians and international participants with experience in dealing with multiple languages, including Dr Antoine Pironet from the Belgian Cancer Registry who presented their work on processing data from bilingual pathology reports.

The panel was organised by Gareth Morlais from the Welsh Government and the members of the CorCenCC team, Dawn Knight, Laura Arman and Irena Spasić. Cardiff University is part of the CorCenCC project (Corpws Cenedlaethol Cymraeg Cyfoes – the National Corpus of Contemporary Welsh), which is creating a major language resource for Welsh speakers, Welsh learners, Welsh language researchers and anyone interested in the Welsh language. CorCenCC is a community-driven project, and Welsh speakers from all kinds of backgrounds and of all abilities can get involved.

Dawn Knight, from the CorCenCC team, said: "Due to privacy concerns we are currently not collecting healthcare narratives, but CorcenCC's data collection and management infrastructure can be easily deployed in a secure environment to support the development of text mining tools that could facilitate the provision of healthcare services in Welsh."

Irena Spasić, Co-founder of the HealTex network, added: "We have been developing tools and resources that can automate the processing of Welsh text. Our software FlexiTerm can extract domain-specific terminology from Welsh documents on the fly. Our early experiments with parallel corpora demonstrated that domain-specific terminologies can be extracted automatically and mapped between Welsh and English. Such functionality can improve the performance of machine translation of specialised texts such as those found in healthcare."

The conference was sponsored by EPSRC via the UK Healthcare Text Analytics Network, the Connected Health Cities, the SAIL databank and the Data Innovation Research Institute at Cardiff University, as well as Averbis Text Analytics and DeepCongito.

The next conference will be organised by Kings College London in 2020.