Publicly Accessible Large Language Model Responses to Frequently Asked Questions About Spondylodiscitis: Preliminary Expert Evaluation - Summary - MDSpire
Advertisement
Evaluation of Responses from Publicly Available Large Language Models to Common Inquiries Regarding Spondylodiscitis: Initial Expert Assessment
To evaluate spine surgeons’ ratings of responses generated by large language models (LLMs) to frequently asked questions about spondylodiscitis.
Approach:
Identification of Relevant FAQs: A chronological, multisource workflow was applied to identify patient-oriented FAQs about spondylodiscitis through Google and PubMed searches.
Key Findings:
Spondylodiscitis is a rare but increasingly common condition with high morbidity and an in-hospital mortality rate of 17.2% within the first year after diagnosis.
Patients often seek additional information online, where LLMs like ChatGPT and Google Gemini can provide accessible health information.
The performance of LLMs in generating responses specific to spondylodiscitis has not been systematically evaluated prior to this study.
Interpretation:
The study highlights the need for evaluating LLMs in the context of complex medical conditions like spondylodiscitis to ensure the accuracy and reliability of the information provided.
Limitations:
The study is preliminary and focuses on a limited set of FAQs.
Responses were rated by spine surgeons, which may not fully represent patient perspectives.
Conclusion:
This evaluation serves as an initial step in assessing the utility of LLMs for providing information on spondylodiscitis.
by Melanie Ardelt, David Schiffelholz, Siegmund Lang, Josina Straub, Sonja Häckel, Nicolas von der Hoeh, Marc Dreimann, Jonathan Neuhoff, Sebastian Siller, Denis Bratelj, Volker Alt, Dietmar Dammerer, Jonas Krueckel
With an aging population, spine disorders are becoming increasingly common. Age-related spinal degeneration is nearly universal, but not all patients experience symptoms—and not all degeneration progresses the same way.
