Automated algorithm for predicting the risk of refractory postpartum hemorrhage

Postpartum hemorrhage (PPH) remains a significant factor in maternal mortality and morbidity worldwide. Fatal outcomes associated with PPH can be potentially prevented through effective prediction and prevention. Methods for PPH prevention have been developed, regulated by clinical guidelines and have found wide application in most countries of the world. However, to date, there is no effective system for identifying patients with a high risk of PPH who require more stringent and scientifically based preventive measures. Aim. To develop and evaluate the informativeness of a computer program (CP) for predicting the risk of refractory PPH based on anamnestic, clinical and laboratory parameters. Materials and methods. Data processing and model building were performed using Python 3.12 and pandas, shap, xgboost, sklearn and mlxtend libraries. Ensemble extreme gradient boosting (XGBoost) models were trained on the selected features. The SHAP method was used to estimate the contribution of each feature to the predictive ability of the models, visualized in bar charts and bee swarm graphs. The developed models were tested on an independent sample of 556 women (the study design was a continuous cross-sectional onetime study). Results. As a result of the conducted study using the available databases, 9 clinical and anamnestic (patient age, age at menarche, parity of delivery, uterine scar, emergency cesarean section, one paraclinical (placenta localization on the anterior wall of the uterus according to ultrasound examination data) and four laboratory (HB, Ht, APTT, fibrinogen levels) parameters were selected from 178 parameters. They were used as the basis for two automated models of the CP for the computer "Prediction of the risk of postpartum hemorrhage". In the model based on the assessment of clinical and anamnestic parameters, the most significant were the presence of a scar on the uterus and the localization of the placenta on the anterior wall of the uterus. In the model based on the assessment of clinical and laboratory parameters, the most important were the levels of Hb and Ht. Conclusion. Two sufficiently informative models of the program "Prediction of the risk of refractory postpartum hemorrhage" have been developed, based on the assessment of clinical and anamnestic (AUC – 0.69) and clinical and laboratory data (AUC – 0.74), the use of which can contribute to the correct stratification of patients in the high-risk group for PPH for the purpose of a more differentiated approach to preventive measures.

1. Mathewlynn S. J., Soltaninejad M., Collins S. L. Artificial Intelligence and Postpartum Hemorrhage. Matern. Fetal Med. 2025;7(1):22–28. https://doi.org/10.1097/FM9.0000000000000257

2. Liu L.Y., Nathan L., Sheen J. J., Goffman D. Review of current insights and therapeutic approaches for the treatment of refractory postpartum hemorrhage. Int. J. Womens Health. 2023;15:905–926. https://doi.org/10.2147/IJWH.S366675

3. Artymuk NV, Artymuk DA, Apresyan SV, Marochko T YU, Atalyan AV, Shibelgut NM, et al. Clinical and anamnestic features of patients with refractory postpartum hemorrhage. Obstetrics and gynecology. News. Views. Education. 2025;13(S):69–76. https://doi.org/10.33029/2303-9698-2025-13-suppl-69-76

4. Andrikopoulou M., D’Alton M. E. Postpartum hemorrhage: early identification challenges. Semin. Perinatol. 2019;43(1):11–17. https://doi.org/10.1053/j.semperi.2018.11.003

5. Klinicheskie rekomendacii – Poslerodovoe krovotechenie – 2021-2022-2023 (16.01.2023) – Utverzhdeny Minzdravom RF. Moscow, 2021. Aviable on: http://disuria.ru/_ld/12/1271_kr21O67O72MZ.pdf. Accessed: Accessed: 28.09.2025.

6. Escobar M. F., Nassar A. H., Theron G., Barnea E. R., Nicholson W., Ramasauskaite D., et al. FIGO recommendations on the management of postpartum hemorrhage 2022. Int. J. Gynaecol. Obstet. 2022;157(suppl. 1):3–50. https://doi.org/10.1002/ijgo.14116

7. World Health Organization. WHO recommendations for the prevention and treatment of postpartum haemorrhage. Ссылка активна на 28.09.2025.https://www.who.int/reproductivehealth/publications/maternal_perinatal_health/9789241548502/en/

8. Prevention and Management of Postpartum Haemorrhage: Greentop Guideline No 52. BJOG. 2017;124(5):e106–e149. https://doi.org/10.1111/1471-0528.14178

9. Semrau K., Litman E., Molina R. L., Marx Delaney M., Choi L., Robertson L., et al. Implementation strategies for WHO guidelines to prevent, detect, and treat postpartum hemorrhage. Cochrane Database Syst. Rev. 2025;2(2):CD016223. https://doi.org/10.1002/14651858.CD016223

10. Ng'ang'a J., Chitimbe T., Mburu R., Rushwan S., Ntirushwa D., Chinery L. et al. Challenges in updating national guidelines and essential medicines lists in Sub-Saharan African countries to include WHOrecommended postpartum hemorrhage medicines. Int. J. Gynaecol. Obstet. 2022;158 Suppl 1:11–13. https://doi.org/ 10.1002/ijgo.14269

11. Holcroft S., Karangwa I., Little F., Behoor J., Bazirete O. Predictive Modelling of Postpartum Haemorrhage Using Early Risk Factors: A Comparative Analysis of Statistical and Machine Learning Models. Int. J. Environ Res. Public. Health. 2024;21(5):600. https://doi.org/10.3390/ijerph21050600

12. Artymuk NV, Marochko TYu, Artymuk DA, Apresyan SV, Kolesnikova NB, Atalyan AV, et al. Risk factors and protective factors for refractory postpartum hemorrhage. Akusherstvo i Ginekologiya/ Obstetrics and Gynecology. 2024;(10):82–90. (in Russ.). https://dx.doi.org/10.18565/aig.2024.169

13. Omotayo M.O., Abioye A.I., Kuyebi M., Eke A.C. Prenatal anemia and postpartum hemorrhage risk: A systematic review and metaanalysis. J. Obstet. Gynaecol. Res. 2021;47(8):2565–2576. https://doi.org/10.1111/jog.14834

14. Ende H. B., Lozada M. J., Chestnut D. H., Osmundson S.S., Walden R.L., Shotwell M. S., et al. Risk factors for atonic postpartum hemorrhage: a systematic review and meta-analysis. Obstet. Gynecol. 2021;137(2):305–323. https://doi.org/10.1097/AOG.0000000000004228

15. Patek K., Friedman P. Postpartum Hemorrhage-Epidemiology, Risk Factors, and Causes. Clin. Obstet. Gynecol. 2023;66(2):344–356. https://doi.org/10.1097/GRF.0000000000000782

16. Artymuk DA, Artymuk NV, Marochko TY, Atalyan AV, Shibelgut NM, Batina NA, et al. Basic parameters of hemostasis in patients with refractory postpartum hemorrhage. Fundamental and Clinical Medicine. 2025;10(2):8–14. (In Russ.). https://doi.org/10.23946/2500-0764-2025-10-2-44-55

17. Yunas I., Islam M. A., Sindhu K. N., Devall A. J., Podesek M., Alam S. S., et al. Causes of and risk factors for postpartum haemorrhage: a systematic review and meta-analysis. Lancet. 2025;405(10488):1468–1480. https://doi.org/10.1016/S0140-6736(25)00448-9

18. Lagrew D M. J., Sakowski C., Cape V., McCormick E., Morton C. H., Lagrew D., et al. Improving Health Care Response to Obstetric Hemorrhage, a California Maternal Quality Care Collaborative Toolkit, 2022. 2022. Ссылка активна на 28.09.2025. https://www.cmqcc.org/resources-tool-kits/toolkits/ob-hemorrhage-toolkit.

19. Susanu C., Hгrгbor A., Vasilache I. A., Harabor V., Călin A. M. Predicting Intra- and Postpartum Hemorrhage through Artificial Intelligence. Medicina (Kaunas). 2024;60(10):1604. https://doi.org/10.3390/medicina60101604

20. Shah S. Y., Saxena S., Rani S. P., Nelaturi N., Gill S., Tippett Barr B., et al. Prediction of postpartum hemorrhage (PPH) using machine learning algorithms in a Kenyan population. Front. Glob. Womens Health. 2023;4:1161157. https://doi.org/10.3389/fgwh.2023.1161157

21. Zhang Y., Chen L., Zhou W., Lin J., Wen H. Nomogram to predict postpartum hemorrhage in cesarean delivery for twin pregnancies: a retrospective cohort study in China. Front. Med. (Lausanne). 2023;10:1139430. https://doi.org/10.3389/fmed.2023.1139430

22. Venkatesh K. K., Strauss R. A., Grotegut C. A., Heine R. P., Chescheir N. C., Stringer J. S. A., et al. Machine Learning and Statistical Models to Predict Postpartum Hemorrhage. Obstet. Gynecol. 2020;135(4):935–944. https://doi.org/10.1097/AOG.0000000000003759

23. Mehrnoush V., Ranjbar A., Farashah M.V., Darsareh F., Shekari M., Jahromi M. S. Prediction of postpartum hemorrhage using traditional statistical analysis and a machine learning approach. AJOG Glob. Rep. 2023;3(2):100185. https://doi.org/10.1016/j.xagr.2023.100185

24. Liu J., Wang C., Yan R., Lu Y., Bai J., Wang H., et al. Machine learning-based prediction of postpartum hemorrhage after vaginal delivery: combining bleeding high risk factors and uterine contraction curve. Arch. Gynecol. Obstet. 2022;306(4):1015–1025. https://doi.org/10.1007/s00404-021-06377-0

25. Obermeyer Z., Emanuel E. J. Predicting the Future–Big Data, Machine Learning, and Clinical Medicine. N. Engl. J. Med. 2016;375:1216–1219. https://doi.org/10.1056/NEJMp1606181

26. Sanchez-Pinto L. N., Luo Y., Churpek M. M. Big Data and Data Science in Critical Care. Chest. 2018;154:1239–1248. https://doi.org/10.1016/j.chest.2018.04.037

27. Savin M. L., Mihai F., Gheorghe L., Lupascu Ursulescu C., Negru D., Trofin A.M., et al. Proposal of a Preoperative CT-Based Score to Predict the Risk of Clinically Relevant Pancreatic Fistula after Cephalic Pancreatoduodenectomy. Medicina (Kaunas). 2021;57(7):650. https://doi.org/10.3390/medicina57070650