Journals / Hitit Medical Journal / 2026 / Cilt: 8 Sayı: 2

A Comparative Assessment of an AI-Based Decision Support Tool and Clinical Evaluation in Predicting Delivery Mode and Timing

Doğum Şekli ve Zamanının Öngörülmesinde Yapay Zekâ Tabanlı Bir Karar Destek Aracının Deneyimli Bir Obstetrisyenin Klinik Değerlendirmesi ile Karşılaştırılması

Seniye Burcu Torumtay Alıç(Hitit Üniversitesi Tıp fakültesi, Kadın Hastalıkları ve Doğum Anabilim Dalı)0009-0007-0659-4726
Pages
272–282
Publication Date
2026-06-29
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Integration of real-time, multimodal intrapartum data and the development of obstetrics-specific AI models may enhance clinical applicability.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Objective: T This study aimed to compare the predictive performance of ChatGPT-4.0, a general-purpose artificial intelligence (AI) language model, with that of an experienced obstetrician in predicting the mode and timing of delivery in low-risk term pregnancies.
📊
Key Findings & Data:Methods: This single-center, retrospective study included 50 low-risk, term, singleton pregnancies.
💡
Conclusion & Impact:Integration of real-time, multimodal intrapartum data and the development of obstetrics-specific AI models may enhance clinical applicability.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Yapay zekâ · Doğum · Sezaryen

Abstract

Objective: T This study aimed to compare the predictive performance of ChatGPT-4.0, a general-purpose artificial intelligence (AI) language model, with that of an experienced obstetrician in predicting the mode and timing of delivery in low-risk term pregnancies. Methods: This single-center, retrospective study included 50 low-risk, term, singleton pregnancies. Exclusion criteria were a history of cesarean section (CS), multiple gestations, fetal macrosomia, maternal height <150 cm, high body mass index, pelvic disproportion, comorbidities, high-risk pregnancy, fetal distress, labor induction, or epidural analgesia. All fetuses were in cephalic presentation, and all participants expressed willingness for vaginal delivery (VD). Demographic and obstetric characteristics, along with admission cardiotocography (CTG) findings, were recorded. Both an experienced obstetrician and ChatGPT-4.0 predicted the mode of delivery and, for VD’s, estimated the timing of birth. ChatGPT-4.0 used CTG interpretations based on consensus among three independent clinicians, whereas the obstetrician interpreted CTGs independently. Statistical analyses included McNemar’s test, Cohen’s kappa coefficient, the Wilcoxon signed-rank test, and agreement analyses. Results: Of the participants, 41 (82%) delivered VD and 9 (18%) underwent CS due to arrest of labor. ChatGPT-4.0 correctly predicted 100% of CS cases, while its accuracy for vaginal deliveries was 53.7%. The obstetrician achieved 100% accuracy for VD’s but only 22.2% for CS. McNemar’s test demonstrated the superiority of ChatGPT-4.0 in predicting delivery mode (p<0.001). Subgroup analysis showed a significant advantage for ChatGPT-4.0 among nulliparous women (p=0.031). No significant difference was found between the two methods in predicting delivery timing (p=0.319). Conclusion: Although not specifically designed for obstetric use, ChatGPT-4.0 demonstrated a notable ability to identify cases requiring CS more accurately than an experienced obstetrician. However, its accuracy in predicting VD remains limited. Integration of real-time, multimodal intrapartum data and the development of obstetrics-specific AI models may enhance clinical applicability.

Özet

Amaç: Bu çalışmanın amacı, düşük riskli term gebeliklerde doğum şekli ve doğum zamanının öngörülmesinde, genel amaçlı bir yapay zekâ (YZ) dil modeli olan ChatGPT-4.0’ın öngörü performansını deneyimli bir obstetrisyenin klinik değerlendirmesi ile karşılaştırmaktır. Gereç ve Yöntemler: Bu tek merkezli, retrospektif çalışmaya düşük riskli, term, tekil gebeliği olan 50 olgu dahil edilmiştir. Önceki sezaryen öyküsü, çoğul gebelik, fetal makrozomi, maternal boy <150 cm, yüksek vücut kitle indeksi, pelvik uygunsuzluk, eşlik eden komorbiditeler, yüksek riskli gebelik, fetal distres, doğum indüksiyonu veya epidural analjezi uygulanan olgular çalışma dışı bırakılmıştır. Tüm fetüsler sefalik prezentasyonda olup olguların tamamı vajinal doğum (VD) isteğine sahipti. Demografik ve obstetrik veriler ile başvuru anındaki kardiyotokografi (KTG) bulguları kaydedildi. Deneyimli obstetrisyen ve ChatGPT-4.0, doğum şeklini ve VD’lerde doğum zamanını tahmin etti. ChatGPT, üç bağımsız klinisyenin uzlaşı ile belirlediği KTG yorumlarını kullanırken, obstetrisyen KTG’leri bağımsız olarak değerlendirdi. İstatistiksel analizlerde McNemar testi, Cohen kappa katsayısı, Wilcoxon işaretli sıralar testi ve uyum analizleri kullanıldı. Bulgular: Olguların 41’i (%82) VD, 9’u (%18) eylem arresti nedeniyle sezaryen ile doğum yaptı. ChatGPT sezaryen olgularının %100’ünü doğru tahmin ederken, vajinal doğumları tahmin etme doğruluğu %53,7 idi. Obstetrisyen VD’lerin %100’ünü doğru tahmin etmiş, ancak sezaryen tahmin doğruluğu %22,2 olarak saptanmıştır. McNemar testi ChatGPT’nin doğum şekli tahmininde üstün olduğunu göstermiştir (p<0,001). Nullipar olgularda ChatGPT lehine anlamlı fark bulunmuştur (p=0,031). Doğum zamanı tahmininde iki yöntem arasında anlamlı fark saptanmamıştır (p=0,319). Sonuç: ChatGPT-4.0, obstetriye özgü bir model olmamasına rağmen sezaryen gereksinimi olabilecek olguları deneyimli bir obstetrisyene kıyasla daha doğru tanımlayabilmiştir. Ancak VD öngörüsündeki doğruluğu sınırlıdır. Gerçek zamanlı ve çok modlu intrapartum verilerle geliştirilecek obstetriye özgü YZ modelleri klinik kullanım potansiyelini artırabilir.

🔖

Cite This Article

Alıç, S. B. T. (2026). A Comparative Assessment of an AI-Based Decision Support Tool and Clinical Evaluation in Predicting Delivery Mode and Timing. Hitit Medical Journal, Cilt: 8 Sayı: 2, 272–282. https://doi.org/10.52827/hititmedj.1846008

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (15)

  1. Yaseen I, Rather RA. A theoretical exploration of artificial intelligence’s impact on feto-maternal health from conception to delivery. Int J Womens Health. 2024;16:903-915.
  2. Guedalia J, Lipschuetz M, Cohen SM, et al. Transporting an artificial intelligence model to predict emergency cesarean delivery: overcoming challenges posed by interfacility variation. J Med Internet Res. 2021;23(12):e28120.
  3. Wie JH, Lee SJ, Choi SK, et al. Prediction of emergency cesarean section using machine learning methods: development and external validation of a nationwide multicenter dataset in the Republic of Korea. Life (Basel). 2022;12(4):604.
  4. Gunnarsson B, Skogvoll E, Jónsdóttir IH, et al. On predicting time to completion for the first stage of spontaneous labor at term in multiparous women. BMC Pregnancy Childbirth. 2017;17(1):183.
  5. Aeberhard JL, Radan AP, Delgado-Gonzalo R, et al. Artificial intelligence and machine learning in cardiotocography: a scoping review. Eur J Obstet Gynecol Reprod Biol. 2023;281:54-62.
  6. Michalitsi K, Metallinou D, Diamanti A, et al. Artificial intelligence in predicting the mode of delivery: a systematic review. Cureus. 2024;16(9):e69115.
  7. Guedalia J, Lipschuetz M, Novoselsky-Persky M, et al. Real-time data analysis using a machine learning model significantly improves prediction of successful vaginal deliveries. Am J Obstet Gynecol. 2020;223(3):437.e1-437.e10.
  8. Khudhur YS. Artificial intelligence in obstetrics and gynecology: current applications and future perspectives. Int J Obstet Gynaecol. 2025;7(1):12-25.
  9. World Health Organization Human Reproduction Programme. WHO statement on caesarean section rates. Reprod Health Matters. 2015;23(45):149-150.
  10. Lipschuetz M, Guedalia J, Cohen SM, et al. Unplanned cesarean delivery in the second stage of labor holds higher odds of complications than in the first stage, while similar in primiparas and multiparas. Fetal Diagn Ther. 2020;47(7):565-571.
  11. Nesheim BI. Duration of labor: an analysis of influencing factors. Acta Obstet Gynecol Scand. 1988;67(2):121-124.
  12. Neal JL, Lowe NK, Ahijevych KL, et al. Active labor duration and dilation rates among low-risk nulliparous women with spontaneous labor onset: a systematic review. J Midwifery Womens Health. 2010;55(4):308-318.
  13. Rosli AA, Nawi AM, Atan IK, et al. Cervical dilatation at diagnosis of active phase of labour determines the mode of delivery and peripartum outcomes: a retrospective study in a single tertiary centre in Malaysia. BMC Pregnancy Childbirth. 2023;23(1):221.
  14. Gumilar KE, Wardhana MP, Akbar MIA, et al. Artificial intelligence-large language models (AI-LLMs) for reliable and accurate cardiotocography interpretation in obstetric practice. Comput Struct Biotechnol J. 2025;27:1140-1147.
  15. Kebede TN, Azanaw G, Mihret MS. Incidence and predictors of dystocia during the active first stage of labor among mothers at Debre Markos Comprehensive Specialized Hospital, Amhara, Northwest Ethiopia. BMC Pregnancy Childbirth. 2025;25(1):1116.