Evaluating ChatGPT for converting clinic letters into patient-friendly language: a quantitative study

Abstract

BACKGROUND: Previous research has shown that communication with patients in language they understand leads to greater comprehension of treatment and diagnoses, but can be time consuming for clinicians. AIM: We sought to investigate the utility of ChatGPT-4 Classic in translating clinic letters into language patients understood without loss of clinical information, and to assess what impact this had on patients' understanding of letter content. DESIGN & SETTING: Single-blinded quantitative study using objective and subjective analysis of language complexity. METHOD: Twenty-three clinic letters were provided by consultants across eight specialties. Letters were inputted into ChatGPT-4 Classic with a prompt related to improve understanding for patients. Patient representatives were then asked to rate their understanding of the content of letters. RESULTS: Translation of letters by ChatGPT-4 Classic resulted in no loss of clinical information, but did result in significant increase in understanding, satisfaction, and decrease in the need to obtain medical help to translate the letter contents by patient representatives compared with clinician-written originals. CONCLUSION: Overall, we concluded that ChatGPT-4 Classic can be used to translate clinic letters into patient-friendly language without loss of clinical content and that these letters are preferred by patients.

Ähnliche Arbeiten