Akademik Veri Yönetim Sistemi
8. Uluslararası 19. Ulusal Hemşirelik Kongresi, Ankara, Türkiye, 25 - 28 Eylül 2024, ss.627-628
REVIEW OF THE EFFECT OF ARTIFICIAL INTELLIGENCE APPLICATIONS ON DETERMINING THE RISK OF POSTPARTUM HEMORRHAGE
The aim of this study is to systematically review research articles that demonstrate the impact of artificial intelligence applications in determining the risk of postpartum hemorrhage (PPH). The literature search of study was conducted between August 23-28, 2024 using the keywords (“postpartum hemorrhage” or “postpartum haemorrhage”) and “artificial intelligence” in ScienceDirect, PubMed, Web of Science and Scopus databases with advanced search option. After reviewing titles, keywords, and abstracts, twenty English research articles published between 2020-2024, suitable the inclusion criteria and with accessible full texts, were selected for the study. In this systematic review, it was seen that the most commonly AI application in PPH is machine learning (ML) and among the ML models; logistic regression (LR), extreme gradient boosting (XGB), random forest (RF), decision tree (DT) model are the most frequently used methods. Using these methods, the risk of PPH was predicted, at-risk groups and key predictors of PPH were identified, the superiority of ML models was assessed, and comparisons with traditional methods were made. The studies reviewed covered various groups, including vaginal cesarean deliveries, singleton-twin pregnancies, as well as different time periods like prenatal, intrapartum, and postpartum. ML models for PPH risk assessment were found to be more effective and advantageous compared to traditional screening methods. The key predictors of PPH include maternal age, hemoglobin levels at delivery, pregnancy anemia, delivery type, gestational age, perineal laceration, anesthesia, diabetes, gestational hypertension, multiple pregnancy, abnormal placentation, fetal macrosomia, oxytocin use in delivery, in vitro fertilization, and primiparous delivery. Critical indicators for predicting PPH are hemoglobin, white blood cells, platelets, fibrinogen degradation products, calcium, sodium, and potassium. In conclusion, ML models for predicting PPH risk, identifying key predictors and risk groups could significantly reduce maternal mortality rates globally. Further studies with spesific groups and expanded use of AI in clinical practice are necessary.
Keywords: "postpartum hemorrhage" "artificial intelligence"