@article{article_1762169, title={A New Ally in HPV and Cervical Cancer Screening: Age-Tailored Scenario-Based Analysis of ChatGPT}, journal={Türk Jinekolojik Onkoloji Dergisi}, volume={25}, pages={24–32}, year={2025}, author={Akdemir, Celal}, keywords={ChatGPT, Büyük Dil Modelleri, HPV, Serviks Kanseri Taraması, Yaşa Duyarlı Bilgi, Senaryo Bazlı Analiz, Uzman Değerlendirmesi, Sağlık Eğitimi, Halk Sağlığı İletişimi, Klinik Kılavuz Uyumu}, abstract={Introduction: In recent years, large language models such as ChatGPT have gained increasing attention in the field of health education. However, their reliability in providing age-sensitive, accurate, and guideline-compliant information on human papillomavirus (HPV) and cervical cancer screening has not been sufficiently investigated. This study aimed to evaluate the performance of ChatGPT-4 in terms of informativeness and guideline compliance when responding to HPV and cervical screening questions tailored to different age scenarios
Methods: Thirty questions were developed based on three age scenarios (18, 30, and 45 years). Each question was submitted to the June 2025 version of ChatGPT-4. The responses were independently evaluated by a five-member panel consisting of three gynecologic oncology surgeons, one infectious diseases specialist, and one public health specialist. Evaluation was based on four criteria: scientific accuracy, clinical guideline compliance, comprehensibility (ease of understanding), and public health reliability. The term “comprehensibility” was used consistently throughout the study instead of “clarity”. Each criterion was rated on a 5-point Likert scale.
Results: The overall mean score across all criteria was 4.19 ± 0.51. The highest mean score was for guideline consistency (4.22 ± 0.48), followed by public health reliability (4.20 ± 0.54), scientific accuracy (4.19 ± 0.50), and comprehensibility (4.16 ± 0.53). The 30-year-old scenario received the highest overall scores, particularly for scientific accuracy (4.34) and guideline consistency (4.26). The 18-year-old scenario scored highest in comprehensibility (4.28) but slightly lower in public health reliability (4.12). The 45-year-old scenario achieved the highest public health reliability score (4.32) but had marginally lower ratings for scientific accuracy (4.16) and comprehensibility (4.10). Expert comments highlighted ChatGPT’s strengths in health communication and combating misinformation, while pointing out the lack of clinical details and explicit guideline references in some responses.
Conclusion: ChatGPT-4 appears to be an effective tool for promoting HPV vaccination and providing public health information, particularly in younger age groups. However, due to its limitations in clinical decision-making and guideline-based content, its use in patient education should be accompanied by expert oversight. Further research should encompass different model versions, additional evaluation metrics, and user perspectives.}, number={2}, publisher={Türk Jinekolojik Onkoloji Derneği}