ChatGPT in Clinical Rheumatology

Abstract

Abstract Background: Recently, it has been demonstrated that chat generative pre-trained transformer (ChatGPT) has successfully completed the United States Medical Licensing Examination and the Saudi Medical License Exam. We conducted an evaluation of ChatGPT (released on February 13, 2023) using a standard clinical toxicological case involving acute organophosphate poisoning. The performance of ChatGPT in addressing all of our inquiries was satisfactory in general but with some limitations. Context: This study is conducted in the context of identifying and illustrating the benefits and drawbacks of using language models and artificial intelligence (AI) in the medical field, especially rheumatology. Aims: We aimed to illustrate the benefits and limitations of applying the AI and language model in the context of medicine and how it could be valuable, especially in the aspects of diagnosis and treatment of rheumatological diseases. Settings and Design: We wrote a typical and usual presentation for two commonly encountered rheumatological diseases, and we asked ChatGPT four major questions regarding the diagnosis and management. Materials and Methods: We input our typical cases to ChatGPT, and we asked four major questions regarding the diagnosis and management, and then we discussed about how ChatGPT approached it and what kind of limitations or drawbacks are encountered. Statistical Analysis Used: Not applicable. Results: Any practitioner in the field is less likely to overlook the typical, straightforward, and uncomplicated clinical case examples we presented. ChatGPT handled all of our inquiries well, and both the initial and regenerated responses were satisfactory and provided coherent explanations of the underlying logic. Nevertheless, the crucial issue in reality is not about obtaining an accurate diagnosis but rather about taking a suitable medical history and being capable of identifying and confirming the correct signs and symptoms. Conclusion: Language models offer various applications and benefits in the field of rheumatology, such as aiding in medical diagnosis and decision-making, facilitating patient communication and education, and enhancing medical education and training.

Ähnliche Arbeiten