Background: The emergence of sophisticated language models, such as Chat Generative Pre-Trained Transformer (ChatGPT), has generated curiosity across diverse domains, including the field of medicine. Developed by OpenAI, ChatGPT is an artificial intelligence (AI) -driven chatbot utilizing Language Processing Models.

Objectives: The aim of this study was to explore the use of ChatGPT by the Tunisian physicians and more specifically the Tunisian rheumatologists.

Methods: This cross-sectional study (online questionnaire survey) was conducted over a 2-week period amongst Rheumatologists and other specialty clinicians. An online questionnaire was developed, with multiple choice, using the Google Forms platform

Results: The survey was completed by 174 physicians. The predominant responses were received from the family doctors (n=43, 24.7%) followed by rheumatologists (n=35, 20.1%). The mean age of participants was 30.29 ± 3.88 years [24-49], with an average experience of 4.28 ± 3 [1-23] years. Among them, 100 (57.5 %) were residents in training, 45 (25.9) were university hospital doctors,16 (9.2%) were public health doctors and 13 (7.5%) worked in the private sector.

The majority of clinicians (n=120, 69%) admitted being aware of ChatGPT and used it in their professional practice, including internal medicine (n=9, 75%), family medicine (n=36, 84%), and psychiatrists (n=8, 89%).

ChatGPT was used by physicians for medical writing, academic purposes, and clinical domain in 51%(n=88), 10% (n=18) and 13% (n=23) of the cases, respectively.

Most of rheumatologists (n=23, 65%) used ChatGPT and some hesitated to use it due to factors like not feeling the need (n=6), lacking experience (n=3), and concerns about data reliability (n=4).

For those using AI, 17 admitted to employing ChatGPT for medical writing, and thus for grammar and spelling correction (n=9), translation (n=11), idea generation (n=8), paraphrasing (n=12), reference generation (n=2), results interpretation (n=1), and to enhance general writing style (n=17).

Only one person utilized ChatGPT for academic activities, primarily for evaluating pedagogical techniques and preparing medical teaching courses. In clinical activities related to diagnosis, monitoring, or therapy, only four rheumatologists used ChatGPT. Two considered it for etiological discussions, two for listing therapeutic alternatives, and one for aiding in follow-up and therapeutic decision-making. The majority, 23 individuals, admitted using ChatGPT with colleagues, 4 with students, 5 with superiors, 2 with scientific work evaluators, and 3 with patients. Respondents expressed concerns about ChatGPT’s reliability (average 1.97 ± 0.9 [0-3]) on the Likert scale. Data confidentiality when using ChatGPT received an average of 2.27 ± 1.4 [0-4], while concerns about potential copyright law violations were evaluated at 2.15 ± 1.4 [0-4] by rheumatologists.

When comparing rheumatologists to other clinicians, we found that they were similar in all other applications, including medical writing, academic domain, and clinical and therapeutic use.

However, the rheumatologist users of ChatGPT were statistically younger (R: 29.11 + 3.7 years vs NR: 30.59 + 3.87 years, p= 0.043) and used it for a longer time (R: 9.57 + 6.8 months vs NR: 7.19 + 4.31 months, p=0.045) than other clinicians.

Conclusion: In this exploration of ChatGPT’s adoption among Tunisian physicians, particularly rheumatologists, we noticed a significant awareness and utilization of this AI-driven tool in medical practices.

REFERENCES: NIL.

Acknowledgements: NIL.

Disclosure of Interests: None declared.