Interpretation Performance in Chinese and Japanese Medical Consultation Scenarios

Fu Jen Catholic University

Status: Not yet enrolling

Conditions: Cardiology; Outpatients; Pulmonology

Treatments: Other: translation software

Study type: Observational

Funder types: Other

Identifiers

NCT06934031

FJUH113419

Details and patient eligibility

About

This study aims to compare the translation and interpretation performance of ChatGPT, Google Translate(GT),and UD Talk in cardiopulmonary consultations and analyze their effectiveness in addressing language barriers.The investigators hypothesize that ChatGPT and UD Talk will outperform GT in terms of accuracy and error rates.

Full description

Background： Language barriers in healthcare have a significant impact on patient safety, health outcomes, and the quality of healthcare services. With the increase in global migration, more patients are facing language challenges in foreign healthcare systems. Literature shows that providing professional interpretation services significantly improves patient satisfaction and communication quality. However, due to the high usage of ad-hoc interpreters and the shortage of professional interpreters, clinical communication quality has declined.

Study Design： This is a one-year, single-center, prospective observational study. Methods： The study will be conducted in the cardiology and pulmonology outpatient clinics at Fu Jen University Hospital. A total of 20 cardiopulmonary disease patients will be enrolled, withtheir consultation sessions recorded and transcribed into Chinese. The study will compare the three tools' performance in terms of semantic accuracy, error types, and severity, and analyze their feasibility in clinical practice. Further analysis will involve satisfaction surveys from both professional interpreters and non-native speakers living in Taiwan. The translation results will be evaluated for accuracy and error rates by 8 professional medical interpreters, with a satisfaction survey completed by a total of 16 experts and non-experts. The evaluation tool will be based on the assessment rubrics from the National Accreditation Authority for Translators and Interpreters.

Effect： It is expected that ChatGPT and UD Talk will show better translation accuracy and interpretation quality compared to GT. UD Talk is anticipated to perform better than the other two tools in real-time interpretation. The study results will provide valuable insights for future medical interpretation training and clinical applications.

Enrollment: 36 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: Accepts Healthy Volunteers

Inclusion and exclusion criteria

First part

Inclusion Criteria:

• Age over 18 years
• Patients visiting the Cardiology or Pulmonology Department at Fu Jen Catholic University Hospital from December 2024 to November 2025

Exclusion Criteria:

• Patients unable to communicate effectively as assessed by the physician
• Refusal to participate in the study

Second part

Inclusion Criteria:

• Age over 18 years
• Completed the Japanese-Chinese medical interpretation training course and has at least two years of clinical experience

Exclusion Criteria:

• Refusal to participate in the study

Third part

Inclusion Criteria:

• Age over 18 years
• Non-native Chinese-speaking Japanese residents who have lived in Taiwan for at least two years

Exclusion Criteria:

• Refusal to participate in the study