Quick Search
A Pilot Study on the Role of Artificial Intelligence in Preterm Labor Prediction
Ece Akca Salık1, Zafer Bütün2, Yeliz Kaya3, Özer Çelik4, Tuğba TAHTA5, Arzu Altın Yavuz61Department of Gynecology and Obstetrics, Eskisehir City Hospital, Eskişehir, Türkiye2Hoşnudiye Mah. Ayşen sokak Dorya rezidans Ablok no: 28/77 Eskişehir, Türkiye
3Eskişehir Osmangazi University Faculty of Health Sciences, Department of Gynecology and Obstetrics Nursing, Eskişehir, Türkiye
4Eskişehir Osmangazi University
5Ankara Medipol University, Health Services Vocational School, Ankara, Türkiye
66Eskişehir Osmangazi University, Faculty of Science, Department of Statistics, Eskisehir, Türkiye
INTRODUCTION: Preterm labor is a major obstetric challenge, contributing significantly to neonatal morbidity and mortality. This study investigates the efficacy of machine learning techniques in predicting the risk of preterm labor among first-time mothers using easily accessible sociodemographic and obstetric information.
METHODS: In this retrospective analysis, data from 102 singleton pregnancies at *** City Hospital, collected between January 2022 and January 2023, was examined. Participants were required to be aged between 18 and 40, have no previous pregnancies, and not suffer from severe medical conditions. The study utilized five machine learning models. The performance of each model was assessed.
RESULTS: Among the models, the XGBoost Classifier achieved the highest accuracy (70%), with a ROC-AUC of 0.64, sensitivity of 60%, and specificity of 80%. BMI and history of abortus imminens were identified as the most significant predictors of preterm labor.
DISCUSSION AND CONCLUSION: Machine learning offers a promising tool for preterm labor prediction based on simple clinical parameters. The findings suggest that BMI and abortus imminens history play a crucial role in risk stratification. Further research with larger, multi-center datasets is necessary to refine predictive accuracy and enhance clinical applicability.
Keywords: Preterm labor, machine learning, artificial intelligence, predictive modeling, nulliparous women.
Manuscript Language: English
