AI in healthcare translation: balancing risk with opportunity

Summary

In an increasingly global healthcare environment, with patients and professionals from many different cultural and linguistic backgrounds, precision in medical document translation is key. Medical documents can range from patient records, patient information leaflets, consent forms, prescriptions, treatment plans to research papers. The translator must have a thorough understanding of the source text and subject matter in order to produce a high-quality target document and ensure patients receive accurate information. But this can come with patient risk, if not done properly.

In this blog, Melanie Cole, Translations Coordinator at EIDO Systems International, talks about the challenges, risks and opportunities for using AI in healthcare translation. 

Content

AI is not a safe replacement

People often ask if I am worried that AI will take over my work, and my answer is always “no”. Professional translators are taught a range of techniques and theory, such as quality assurance, cultural sensitivity, terminology, context, consistency, client preferences, and more importantly, risk mitigation and legal compliance. They are trained in one or several subject areas, such as legal, financial, technical, literary, journalistic, and medical. Medicine, in particular, is a challenging area because of the legal implications and risks involved, so there is often a more specialised approach to training medical translators.

I worry there is a growing misperception that unqualified translators or AI can easily do the job. This is particularly pertinent in healthcare, where there are cases of translated text being presented to patients or clinicians without any form of secondary review. Low-quality translations can have catastrophic consequences, such as inappropriate patient care, incorrect diagnoses or, in extreme cases, death. Unregulated AI translations present a serious risk to patient safety.

Common translations tools

Technology and translation have worked hand in hand for many years. For decades, translators have used a range of computer-assisted translation (CAT) and localisation tools. This is particularly useful for organisations that use the same chunks of text across different publications as it ensures consistency in terminology and style.

It is also common practice for translators to use machine translation engines within their CAT tools, which they review and post-edit. With newer large-language models and neural machine translation, this has advanced even further, and prompt-based AI translation has seen results like never before. However, in my opinion, it is still not good enough for medical translation as it lacks the quality, consistency, style and cultural insight a medical translator can provide. It is also prone to “hallucinations” (where information is made up) and factual errors, which can lead to serious mistranslations.

How can we use AI safely to assist translation in healthcare?

AI should be used to assist translators not replace them. This could be by producing first drafts that are then edited and proofread. It can also support translators with admin-heavy duties, such as terminology and translation memory management and might be useful for informal communication (such as social media) or for getting the “gist” of something in another language. In other words, AI-translated text without human review should never be used for public consumption in healthcare.

At EIDO, we do work with AI where necessary. For example, in India AI has been used to produce a first draft translation of EIDO information in Hindi and Marathi while using our translation memory as the base text. This is then proofread and edited by human translators and sent for a final review by an accredited medical organisation before publishing. This layered review process ensures that any inaccuracies and mistranslations are picked up along the way, and by using a translation memory, translations are kept consistent with previously translated content. This ensures patients in India can be given information leaflets that are both understandable and precise as well as being culturally sensitive to their needs.

How to choose a translation partner

I am often asked how I choose my translation partners. I would suggest using an accredited agency or translator from the ATC, ITI or CIoL. The following factors should be considered when choosing a translator:

• qualifications: ensure your translators are experts in the medical field.
• quality assurance: they should have an adequate process in place (i.e. translation followed by editing, proofreading and review by subject matter experts, where applicable).
• compliance: they should be compliant with healthcare regulations and standards in the target culture and language.
• confidentiality policies: should be in place to protect any sensitive information.

In the case of charities or non-profit organisations with budgeting constraints, I would recommend partnering with non-profit organisations, such as Translators without Borders. Some organisations may still make use of AI, but it must be used wisely, especially in the medical context, and by using accredited translation agencies or organisations, you are assured that your medical translation has all the necessary qualities.

Final thoughts

The use of AI in translation is a controversial subject, particularly in healthcare. There are serious cultural, medical and legal implications arising from the accuracy of information provided to both healthcare professionals and patients. These risks, if unmitigated, could lead to serious patient harm. AI translation without human supervision and review is therefore a false economy. It should not be used in medical translation unless handled by a professionally trained medical translator. By investing in high-quality medical translation, you can rest assured that patients receive accurate, safe information which leads to the right care and treatment.

Opinions expressed in blogs and other content are those of the author. Patient Safety Learning welcomes sharing content and opinions that promotes safer patient care and for the reduction of avoidable harm.  The views expressed on the hub however do not necessarily represent Patient Safety Learning's views or values. References to a specific product or service does not imply a recommendation or endorsement.

About the author

Melanie Cole is the Translations Coordinator at EIDO Systems International. Her job is to coordinate and supervise the translation of a library of over 400 patient information leaflets into over 30 different languages, as and when requested by various partners and customers (hospital trusts, private clinics and individual surgeons).

She qualified as a translator in 2009 and has worked in the industry ever since.