The Role of Interpreters in Supporting Resilience and Implications for Artificial Intelligence

Abstract Background Interpreters help millions of patients with a non-English language preference (NELP) navigate the healthcare system. Artificial intelligence (AI) tools are rapidly improving and undergoing testing and implementation for interpretation purposes. However, little is understood about interpreters’ roles beyond direct interpretation, such as how they may promote resilience, which supports patient safety. Understanding these additional roles is critical as AI tools expand in interpretation. Objective To understand how medical interpreters enhance resilience in the care of patients with NELP undergoing ambulatory diagnostic processes. Design We conducted focus groups with medical interpreters to examine their experiences assisting patients with NELP in ambulatory visits for new or worsening symptoms. We focused on the ambulatory setting due to the high risk of misdiagnosis as patients navigate language barriers and the health system across multiple encounters. Participants Seventeen medical interpreters participated in three focus groups and result-checking. Approach Deductive analysis using Hollnagel’s resilience potentials — potential to respond, monitor, learn, and anticipate — as the organizing framework for understanding the impact of interpreter behaviors. Results Eleven behaviors supporting resilience were identified. The potential to respond included (1) managing communication flow, (2) strengthening the patient’s voice, (3) viewing their interpretation role broadly, and (4) engaging in pre-briefings. The potential to monitor involved monitoring (5) patient emotion and understanding, (6) team dynamics, and (7) the impact of technology. The potential to learn included (8) learning from their cultural background and (9) from their work experiences. The potential to anticipate was comprised of (10) predicting future threats to patient safety and (11) identifying potential system improvements. Conclusions Interpreters engage in behaviors beyond strict interpretation which enhance resilience, but these behaviors are not yet supported by AI interpretation tools. Careful consideration of these resilience-enhancing behaviors will be necessary when implementing AI tools and their workflows to avoid safety gaps.