Is ChatGPT Better Than Epileptologists at Interpreting Seizure Semiology?

Importance: Utilizing large language models (LLMs), primarily ChatGPT, to interpret the seizure semiology with focal epilepsy could yield valuable data for presurgical assessment. Assessing the reliability and comparability of LLM-generated responses with those from well-trained neurologists, especially epileptologists, is crucial for ascertaining the value of LLMs in the presurgical evaluation. Objective: To compare the quality of seizure semiology interpretations generated by ChatGPT with those provided by epileptologists. Design, Setting, and Participants: A total of 865 descriptions of seizure semiology and validated epileptogenic zone (EZ) pairs were derived from 189 public papers. These semiology records were utilized as input of ChatGPT to generate responses on the most likely locations of EZ. Additionally, a panel of 5 epileptologists was recruited to complete an online survey by providing responses on EZ locations based on 100 well-defined semiology records. All responses from ChatGPT and epileptologists were graded for their reliability score (RS) and regional accuracy rate (RAR). Main outcome measures: Evaluation and comparison of responses from ChatGPT and epileptologists at two granularity levels (general region level and specific region level), using the RS and the RAR. Results: In evaluating responses to semiology queries, the highest RARs in each general region from ChatGPT-4.0 were 89.28% for the frontal lobe and 71.39% for the temporal lobe. However, the RAR was lower for the occipital lobe at 46.24%, the parietal lobe at 31.01%, the insular cortex at 8.51%, and the cingulate cortex at 2.78%. Comparatively, the RAR achieved by epileptologists was 82.76% for the frontal lobe, 58.33% for the temporal lobe, 68.42% for the occipital lobe, 50% for the parietal lobe, 60% for the insular cortex, and 28.57% for the cingulate cortex. Conclusions and Relevance: In this study of seizure semiology interpretation, ChatGPT-4.0 outperformed epileptologists in interpreting seizure semiology originating in the frontal and temporal lobes, whereas epileptologists outperformed ChatGPT-4.0 in the occipital and parietal lobes, and significantly outperformed in the insular cortex and cingulate cortex. ChatGPT demonstrates the potential to assist in the preoperative assessment for epilepsy surgery. Presumably, with the continuous development of LLM, the reliability of ChatGPT will be strengthened in the foreseeable future. ### Competing Interest Statement Dr. Aboud has served on the advisory board for Servier and is supported in part by the UC Davis Paul Calabresi Career Development Award for Clinical Oncology as funded by the National Cancer Institute/National Institutes of Health through grant #2K12CA138464-11. Dr. Rosenow has received research support from the Federal State of Hesse, specifically at the Center for Personalized Translational Epilepsy Research from 2018 to 2022. Dr. Rosenow has received research support from Chaja-Foundation Frankfurt, focusing on establishing and evaluating the ketogenic diet in institution. Dr. Rosenow received research support from Reiss-Foundation Frankfurt, mainly for the research on the ketogenic diet in GLUT1-DS. Dr. Rosenow received research support from German Ministry of Education, focusing on the ERAPerMed Raise-Genic. ### Funding Statement This study did not receive any funding。 ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The data used in this paper were collected from published papers from multiple journals and publishers. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present study are available upon reasonable request to the authors after the completion of the project.