Page 65 - AIH-1-3
P. 65

Artificial Intelligence in Health                                       ChatGPT in writing scientific articles




                         A                                  B










                         C                                  D














            Figure 4. Impact factors distribution of sources of articles on the topic “biotelemetry in cardiology” using the first prompt. The Y-axis of the charts presents
            the source quantity. The color fill indicates the source classification: Green: Reliable/semi-reliable; Red: Fictitious. (A and C) correspond to the results of
            ChatGPT 3.5, whereas (B and D) correspond to the results of ChatGPT 4. Image created using MATLAB (R2021b, The MathWorks Inc., USA).

                         A                                  B













            Figure 5. Impact factor distribution of article sources generated by the third prompt for the ChatGPT 4. The Y-axis of the charts presents the source
            quantity. (A) Results of the topic of “biotelemetry in cardiology.” (B) Results of the topic of “biotelemetry in oncology.” Image created using MATLAB
            (R2021b, The MathWorks Inc., USA).

               «Please note that the reference numbers and details   In some instances, ChatGPT added URL links to non-
               provided are illustrative and represent the type of   existent web pages. Similarly, articles with fictitious DOIs
               sources that would be ideal for a paper like this. In an   were periodically generated during the work with both
               actual academic setting, these references would need   versions of ChatGPT. Often, these DOIs either belonged
               to be sourced from real, high-impact journals and   to a real article but with completely different titles, or were
               articles»                                       entirely made up.

              Furthermore, this version incorporates the addition of a   Another characteristic observed was the partial
            note to verify the availability of resources when composing   replacement of the real titles according to the prompt. For
            a scholarly publication:                           example, the word “oncology” might be added to an article
                                                               about biotelemetry, resulting in a new and fictitious article
               «Remember, these references are indicative and should   title. Such additions of topic words were characteristic of
               be used as starting points for more in-depth research.   all topics. For telemedicine in the area of “remote medical
               For the latest studies and reviews, accessing academic   examination,” both versions of ChatGPT often changed
               databases like PubMed, Scopus, or Google Scholar is   the beginning of words, turning existing articles about
               recommended. Always check the publication date to   physiotherapy into fictitious articles about psychotherapy
               ensure the information is current and relevant»  and vice versa.


            Volume 1 Issue 3 (2024)                         59                               doi: 10.36922/aih.2592
   60   61   62   63   64   65   66   67   68   69   70