Letter to Editor Regarding “Assessing the Capability of ChatGPT, Google Bard, and Microsoft Bing in Solving Radiology Case Vignettes”

Assessing the Capability of ChatGPT, Google Bard, and Microsoft Bing in Solving Radiology Case Vignette

We recently read with great interest the article by Sarangi and colleagues, published in the December 2023 issue of the Indian Journal of Radiology and Imaging that assessed ChatGPT, Google Bard, and Microsoft Bing's diagnostic abilities in solving radiology vignettes.[1] The creative application of large language models by the authors holds promise for this field in radiology. We are grateful for their important contribution. The study's overall methodology and conclusions are excellent; however, some methodological issues raised questions in our minds and made us look for further information.

Prompt selection may have an impact on large language models' (LLMs) diagnostic success. Caruccio and colleagues demonstrated how varying prompts, ranging from simple to complex, might alter the chatbots' performance.[2] First, could the authors kindly provide more details on the particular input prompt they used in their research? Second, have the chatbots been trained by the authors?

With its significant implications for integrating LLMs, we sincerely appreciate the chance to discuss this ground-breaking research. We would better grasp the complexities of the study if the previously mentioned points were clarified.

Conflict of Interest

None declared.

Ethical Approval

This article does not contain any studies with human participants or animals performed by any of the authors.

• References

• 1 Sarangi PK, Narayan RK, Mohakud S, Vats A, Sahani D, Mondal H. Assessing the capability of ChatGPT, Google Bard, and Microsoft Bing in solving radiology case vignettes. Indian J Radiol Imaging 2023; (eFirst ahead of print) DOI: 10.1055/s-0043-1777746.
• 2 Caruccio L, Cirillo S, Polese G, Solimando G, Sundaramurthy S, Tortora G. Can ChatGPT provide intelligent diagnoses? A comparative study between predictive models and ChatGPT to define a new medical diagnostic bot. Expert Syst Appl 2024; 235: 121186

Publication History

Article published online:
25 March 2024