| Accuracy of ChatGPT in the diagnosis and treatment of cervical disorders: comparative study to APTA guidelines |
| Paper ID : 1044-ISCSR3 (R2) |
| Authors |
|
Mariam Wagdy Ayoub *1, Marwa El-kawy Ghait Mohamed2, Marwa Mohammed Ahmed3, Mariam Ramdam Hares3 1Student of Physical Therapy at Sphinx University , Assiut 2Assistant lecturer, Department of Physical Therapy for Pediatrics, Faculty of Physical Therapy, Sphinx University, Assiut, Egypt 3Student of Physical Therapy at Sphinx University |
| Abstract |
| ChatGPT is a system that uses a large amount of data , enabling it to understand human language , provide responses for it and perform tasks typically requiring human cognition . It is developing rapidly in all fields of society, including the medical field. Especially in Covid-19, increased need for helping in diagnosis and follow up from home . Objective This paper aims to evaluate efficiency , accuracy and ability of ChatGPT in generating physical therapy treatment plans . Method Thirty-three cases of cervical conditions in their twenties, diagnosed with spondylosis with neck pain and brachialgia, spondylosis with headache, and spondylosis with muscle spasm, were selected from the Physiotherapy Clinic of Sphinx University. The cases were entered into ChatGPT-2, which was tasked with generating physical therapy treatment plans for them. Additionally, treatment plans were obtained from APTA . A comparison was made between the treatment plans generated by ChatGPT-2 and APTA guidelines for each case, by Cohens kappa . Result After comparing the PT treatment plan provided by ChatGPT with APTA guidelines for cervical conditions, Cohens kappa value for agreement between them was 0.5 ,that indicates moderate agreement between ChatGPT recommendation and APTA guidelines . And observed that The treatment plan generated by ChatGPT offers a generalized strategy for all cervical cases, regardless of the specific disease or the nature of the condition (acute, subacute, or chronic). The APTA guidelines categorizes cervical cases into four groups, ensuring evidence-based, condition-specific interventions. In contrast, ChatGPT’s plan includes a broader range of modalities, some lacking strong scientific support . Conclusion This study suggests that physical therapy treatment plans generated by ChatGPT are not sufficiently accurate. While ChatGPT’s plans incorporate APTA recommendations, they also include treatment modalities that are not scientifically proven. Therefore, further research is needed to evaluate its effectiveness |
| Keywords |
| Keywords : cervical , Artificial intelligence , treatment plan , ChatGPT . |
| Status: Abstract Accepted |