Abstract
When faced with language barriers, UK healthcare staff have found themselves turning to machine translation (MT) – predominantly Google Translate – to fulfil their duty of care to patients [3, 16, 23]. Despite the risks potentially posed by the use of MT in such complex and sensitive situations, little research currently exists as to healthcare staff awareness of these risks in real-life settings. This gap is particularly notable concerning the use of MT with patient medical record information compared with interpersonal situations and patient-oriented documentation [6, 7, 8, 14, 16, 22, 27]. While research has been conducted into the perceptions and practices of the general population concerning MT use in largely lower-stakes contexts [31], research on the extent to which these transfer to higher-stakes settings remains lacking. The contribution this paper aims to make is therefore twofold: to investigate the impact of MT on patient medical record documentation and to explore the extent to which healthcare staff are aware of the risks potentially posed by its use. In this paper, we selected contextualised medical abbreviation examples from authoritative French and Spanish clinical corpora [9, 15] to serve as a use case, abbreviations having previously been shown to pose an increased risk for patient harm even prior to their translation with Google Translate [2, 5, 12, 19, 25, 28]. Examples containing higher-risk MT errors were presented to healthcare staff to ascertain their perceptions and risk awareness as part of semi-structured interviews. Whilst these interviews remain ongoing, this paper presents the findings on risks identified in the use of MT with patient medical documentation, and the responses obtained thus far.