Prompting language influences diagnostic reasoning and accuracy of large language models

TL;DR

This study shows that the language used to prompt large language models significantly affects their diagnostic reasoning and accuracy in clinical settings, with English generally yielding better results than French.

Contribution

It provides a comparative analysis of LLM performance across languages, highlighting the importance of prompt language in clinical decision support applications.

Findings

Most models perform better in English than French.

Language impacts reasoning quality and diagnosis accuracy.

o3 model shows no significant language effect.

Abstract

Large language models (LLMs) are increasingly explored for clinical decision support, yet most evaluations are conducted in English, leaving their reliability in other languages uncertain. Here we evaluate the impact of prompting language on diagnostic reasoning and final diagnosis accuracy by comparing English and French performance across five LLMs (o3, DeepSeek-R1, GPT-4-Turbo, Llama-3.1-405B-Instruct, and BioMistral-7B). A total of 180 clinical vignettes covering 16 medical specialties were assessed by two physicians using an 18-point scale evaluating both diagnosis accuracy and reasoning quality. Four of the five models performed better in English (mean difference 0.37-0.91, adjusted p < 0.05), with the gap spanning multiple aspects of reasoning, including differential diagnosis, logical structure, and internal validity. o3 was the only model showing no overall language effect. These findings demonstrate that prompting language remains a critical determinant of LLM clinical performance, with implications for equitable linguistico-cultural deployment worldwide.