Although Google Translate was less than 100 percent accurate in translating emergency exit instructions into English for patients who speak Spanish and Chinese, researchers cautiously support its use, a study published in JAMA Internal Medicine find.
The researchers, at the University of California, San Francisco, analyzed 100 sets of emergency discharge instructions translated by Google’s new machine learning algorithm, released in 2017. They found the algorithm to be accurate 92% for Spanish and 81% for Chinese.
According to the study, only a small percentage of inaccurate translations – 2% in Spanish and 8% in Chinese – had the potential to cause clinically significant harm.
Translation errors were mostly grammatical or typographical in the original English written instructions, and patients who could read English could have understood correctly, the study concluded.
“Google Translate is more accurate than a lot of clinicians believe, and I think it’s definitely more useful than providing nothing at all,” said first study author Elaine Khoong, MD. “We cautiously support its use.”
The algorithm encountered difficulties when doctors used colloquial terms such as “skipping a meal,” a phrase Google translated into Chinese as “skipping” a meal.
A disturbing translation error occurred when a doctor told a patient to “hold the kidney medicine” – that is, to stop taking it. The algorithm translated this in Spanish as “keep the medicine” and in Chinese as “keep taking” the medicine. The researchers called these errors “fatal”.
Machine translation errors were also associated with medical jargon and long, complicated sentences.
The researchers said Google Translate is best used in combination with human interpreters who can interpret the doctor’s verbal instructions for the patient, even over the phone or video, while the doctor shows the patient the written translation.