Giovanni Rodriguez, MD is currently the Clinical Director for Brigham & Women’s Hospital Emergency Department, where she oversees day to day operations and provides care. Prior to this role she was a chief resident in the Harvard Affiliated Emergency Medicine Residency Program at Mass General Brigham in Boston, MA. Dr. Rodriguez has pursued innovative initiatives to address systemic health care disparities particularly for patients with limited English proficiency. Within the field of Emergency Medicine, she aspires to expand access to patients by strengthening programs that enhance the discharge processes, facilitate effective communication, and foster partnerships with community centers. Dr. Rodriguez served as the chair of Residents and Fellows Committee for Center for Diversity and Inclusion at Massachusetts General Hospital. She received the 2023 DEI Resident Education/Innovation Award from the Society of Academic Emergency Medicine. Dr. Rodriguez received her medical degree from the Indiana University School of Medicine in 2016
Giovanni Rodriguez, MD
Clinical Director of Operations, Emergency Department, Brigham and Women's Hospital, Boston, MA
Attending Physician, Emergency Department, Massachusetts General Hospital, Boston, MA
Reimagining Language Access: Evaluating Discharge Practices & AI-Powered Translation Tools
Objectives:
- To assess current language access practices for patients with Limited English Proficiency (LEP) during emergency department discharge.
- To evaluate the accuracy of informal AI translation tools commonly used in clinical settings.
- To identify policy gaps and propose institutional strategies for safer, more equitable use of AI-assisted translation.
Background:
Patients with LEP face significant communication barriers in healthcare, especially at the point of emergency department (ED) discharge, a critical moment for care transitions. Inadequate language access at discharge has been associated with increased adverse events, poor comprehension of care instructions, and inequitable outcomes. Federal mandates under Title VI of the Civil Rights Act standards require meaningful access, yet implementation of written language services remains uneven. Clinicians are increasingly turning to informal tools such as Google Translate and ChatGPT to bridge language gaps, but these tools’ clinical accuracy and safety are largely unvalidated.
The purpose of this practicum project is to evaluate current language access practices for patients with LEP at the point of emergency department discharge and to assess the role of informal AI translation tools in bridging communication gaps. This project aims to identify gaps in institutional policy, evaluate the accuracy and appropriateness of AI-generated translations, and propose actionable strategies to enhance language equity through responsible integration of technology and community-informed solutions.
Methods:
- Conducted a retrospective chart review of 30 ED discharge summaries for LEP patients at Massachusetts General Hospital to assess translation practices and interpreter documentation.
- Evaluated the presence of translated discharge instructions, use of disclaimers, and documentation of professional interpreter involvement.
- Conducted a pilot study comparing the translation accuracy of Google Translate, ChatGPT, and vendor-provided templates across selected ED discharge instructions in Spanish and Chinese.
Results:
Of the discharge summaries reviewed, 50% included translated content; of these, 46% appeared to rely on informal tools such as Google Translate or ChatGPT. Only 1 chart included disclaimers about the limitations of machine-generated translation, and 30% documented the involvement of a professional interpreter at discharge. In our pilot study GPT4 performed comparably to templates for Spanish translations, particularly in fluency and adequacy, but was less consistent for Chinese, where templates outperformed in fluency and meaning. These findings highlight GPT4’s potential to complement traditional translation methods, particularly for Spanish, while emphasizing the need for further study for broader language applications. Importantly, GPT4 translations did not cause harm.
Future Directions:
- Conduct a larger-scale study comparing the performance of ChatGPT and Google Translate in translating ED discharge summaries across Spanish, Chinese, and Portuguese, with assessments conducted by professional medical interpreters.
- Develop policies in for both AMC EDs to improve guidance at discharge.
Preceptor:
Lauren Nentwich, MD- Emergency Department VP of Clinical Affairs, Mass General Brigham
