Chen, C.

Abstract Rationale: Effective communication is crucial in healthcare, especially in acute care settings, so that patients/family members can understand treatment options, and share preferences and goals of care. For those with a non-English language preference (NELP), or who have difficulty communicating in English, professional interpreters are recognized as the gold standard in supporting bidirectional communication. However, interpreters are not always readily available for all languages or clinical settings prompting the exploration of other options for translation and interpretation. The recent developments in artificial intelligence (AI), specifically machine translation have enabled more accurate and emotional machine interpretation and translation. Methods: We conducted a systematic review to explore the usability and usefulness of AI-assisted interpretation and translation tools for improving communication between patients and clinicians including verbal and written communication. A comprehensive search of several databases including Ovid MEDLINE(R), Ovid Embase, APA PsycInfo, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus via Elsevier was performed, resulting in 2629 studies. We included studies evaluating tools for direct patient-clinician communication. For verbal communication, we included articles assessing speech interpretation, and for written communication, we focused on discharge instruction translation. Results: A total of 18 studies were included in the final analysis. Among these, 5 assessed the use of speech interpretation and 13 examined text translation of discharge materials. 14 (78%) studies found these novel AI tools beneficial for patient care, whereas 11 (62%) studies highlighted the risks, indicating that these tools can lead to serious mistakes with potential clinical harm. 5 (28%) studies reported participant satisfaction with the use of these tools in their care. Translation and interpretation accuracy varied notably across languages, with lower accuracy scores observed in languages with limited digital resources, such as Haitian Creole and Vietnamese. Conclusion: AI-assisted interpretation and translation in healthcare has several advantages including broader language accessibility than usual resources and potential cost savings for institutions and demonstrates evidence of patient satisfaction. However, these novel tools operate using multiple phases and each stage is prone to errors which may negatively impact patient care and outcomes.. AI-based interpretation and translation show promising potential, however, it is not yet ready for widespread clinical use, so professional language interpreters should be used whenever possible. Future studies should focus on evaluating accuracy as well as how best to integrate these technologies in real-time clinical settings without compromising patient safety and quality of care.