235. Lost in Translation: Accuracy of Software-Driven Translational Devices in the Pediatric Neurosurgery Clinical Setting

Authors: Michael Cools, MD; Michael Cools, MD; Jorge Feria, MD; Cole Sloboda, BS; Christopher Lintz; Carolyn Quinsey, MD (Durham, NC)

Introduction: Working with pediatric neurosurgical patients and families whose primary language is not English is a significant challenge to delivering effective care. While in-person interpreters are the standard in many settings, they are not available in all settings. Many practitioners have begun using these translation devices to have informal conversations with patients and families. However, the accuracy of these devices in the medical setting has not been tested.

Methods: We compared 3 devices, the Google Pixel buds, Travis Touch, and iTranslate Pro, for accuracy in translation of a list of words and phrases that would commonly be used in a pediatric neurosurgical clinic visit. A native Spanish-speaking physician graded the accuracy as “correct,” “incorrect but understandable,” and “incorrect” for each word and phrase.  The ease of use was also evaluated.

Results: A total of 85 words and phrases were tested for all 3 devices. Pixel buds were the most accurate, with 95% “correct” and only 5% “incorrect.” Travis was next best with 92% “correct” and 7% “incorrect but understandable.” iTranslate was the worst, with only 68% “correct,” 20% “correct but understandable,” and 12% “incorrect.” There was not a significant difference between the Pixel buds and the Travis (p=0.43). However, the iTranslate performed significantly worse than both the Pixel buds (p<0.0001) and the Travis Touch (p=0.0001). The in-ear nature of the Pixel buds allowed for a more natural interaction than the external Travis Touch and iTranslate.

Conclusion: Accuracy of the Travis Touch and Google Translate with Pixel Buds were both good for words and phrases commonly used in a clinical environment. However, iTranslate performed poorly. The use of translation devices, especially with in-ear capabilities, has the potential to enhance the patient-physician interaction in clinical encounters with a language barrier.