Prediction Model for Massive Transfusion in Placenta Previa during Cesarean Section

Interpretable machine learning predicts postpartum hemorrhage with severe maternal morbidity in a lower-risk laboring obstetric population

BACKGROUND

Early identification of patients at increased risk for postpartum hemorrhage (PPH) associated with severe maternal morbidity (SMM) is critical for preparation and preventative intervention. However, prediction is challenging in patients without obvious risk factors for postpartum hemorrhage with severe maternal morbidity. Current tools for hemorrhage risk assessment use lists of risk factors rather than predictive models.

OBJECTIVE

To develop, validate (internally and externally), and compare a machine learning model for predicting PPH associated with SMM against a standard hemorrhage risk assessment tool in a lower risk laboring obstetric population.

STUDY DESIGN

This retrospective cross-sectional study included clinical data from singleton, term births (>=37 weeks' gestation) at 19 US hospitals (2016-2021) using data from 58,023 births at 11 hospitals to train a generalized additive model (GAM) and 27,743 births at 8 held-out hospitals to externally validate the model. The outcome of interest was PPH with severe maternal morbidity (blood transfusion, hysterectomy, vascular embolization, intrauterine balloon tamponade, uterine artery ligation suture, uterine compression suture, or admission to intensive care). Cesarean birth without a trial of vaginal birth and patients with a history of cesarean were excluded. We compared the model performance to that of the California Maternal Quality Care Collaborative (CMQCC) Obstetric Hemorrhage Risk Factor Assessment Screen.

RESULTS

The GAM predicted PPH with an area under the receiver-operating characteristic curve (AUROC) of 0.67 (95% CI 0.64-0.68) on external validation, significantly outperforming the CMQCC risk screen AUROC of 0.52 (95% CI 0.50-0.53). Additionally, the GAM had better sensitivity of 36.9% (95% CI 33.01-41.02) than the CMQCC screen sensitivity of 20.30% (95% CI 17.40-22.52) at the CMQCC screen positive rate of 16.8%. The GAM identified in-vitro fertilization as a risk factor (adjusted OR 1.5; 95% CI 1.2-1.8) and nulliparous births as the highest PPH risk factor (adjusted OR 1.5; 95% CI 1.4-1.6).

CONCLUSION

Our model identified almost twice as many cases of PPH as the CMQCC rules-based approach for the same screen positive rate and identified in-vitro fertilization and first-time births as risk factors for PPH. Adopting predictive models over traditional screens can enhance PPH prediction.

Predicting intraoperative blood loss during cesarean sections based on multi-modal information: a two-center study

PURPOSE

To develop and validate a nomogram model that combines radiomics features, clinical factors, and coagulation function indexes (CFI) to predict intraoperative blood loss (IBL) during cesarean sections, and to explore its application in optimizing perioperative management and reducing maternal morbidity.

METHODS

In this retrospective consecutive series study, a total of 346 patients who underwent magnetic resonance imaging (156 for training and 68 for internal test, center 1; 122 for external test, center 2) were included. IBL+ was defined as more than 1000 mL estimated blood loss during cesarean sections. The prediction models of IBL were developed based on machine-learning algorithms using CFI, radiomics features, and clinical factors. ROC analysis was performed to evaluate the performance for IBL diagnosis.

RESULTS

The support vector machine model incorporating all three modalities achieved an AUC of 0.873 (95% CI 0.769-0.941) and a sensitivity of 1.000 (95% CI 0.846-1.000) in the internal test set, with an AUC of 0.806 (95% CI 0.725-0.872) and a sensitivity of 0.873 (95% CI 0.799-0.922) in the external test set. It was also scored significantly higher than the CFI model (P = 0.035) on the internal test set, and both the CFI (P = 0.002) and radiomics-CFI models (P = 0.007) on the external test set. Additionally, the nomogram constructed based on three modalities achieved an internal testing set AUC of 0.960 (95% CI 0.806-0.999) and an external testing set AUC of 0.869 (95% CI 0.684-0.967) in the pregnant population without a pernicious placenta previa. It is noteworthy that the AUC of the proposed model did not show a statistically significant improvement compared to the Clinical-CFI model in both internal (P = 0.115) and external test sets (P = 0.533).

CONCLUSION

The proposed model demonstrated good performance in predicting intraoperative blood loss (IBL), exhibiting high sensitivity and robust generalizability, with potential applicability to other surgeries such as vaginal delivery and postpartum hysterectomy. However, the performance of the proposed model was not statistically significantly better than that of the Clinical-CFI model.

The relationship between cervical length and area measurements evaluated by MRI and the amount of hemorrhage in PAS cases

BACKGROUND

Placenta accreta spectrum often leads to massive hemorrhage and even maternal shock and death. This study aims to identify whether cervical length and cervical area measured by magnetic resonance imaging correlate with massive hemorrhage in patients with placenta accreta spectrum.

METHODS

The study was conducted at our hospital, and 158 placenta previa patients with placenta accreta spectrum underwent preoperative magnetic resonance imaging examination were included. The cervical length and cervical area were measured and evaluated their ability to identify massive hemorrhage in patients with placenta accreta spectrum.

RESULTS

The cervical length and area in patients with massive hemorrhage were both significantly smaller than those in patients without massive hemorrhage. The results of multivariate analysis show that cervical length and cervical area were significantly associated with massive hemorrhage. In all patients, a negative linear was found between cervical length and amount of blood loss (r =-0.613), and between cervical area and amount of blood loss (r =-0.629). Combined with cervical length and cervical area, the sensitivity, specificity, and the area under the curve for the predictive massive hemorrhage were 88.618%, 90.209%, and 0.890, respectively.

CONCLUSION

The cervical length and area might be used to recognize massive hemorrhage in placenta previa patients with placenta accreta spectrum.

Obstetric blood transfusion in placenta previa patients with prenatal anemia: a retrospective study

BACKGROUND

The appropriate use of obstetric blood transfusion is crucial for patients with placenta previa and prenatal anemia. This retrospective study aims to explore the correlation between prenatal anemia and blood transfusion-related parameters in this population.

METHODS

We retrieved the medical records of consecutive participants who were diagnosed with placenta previa and underwent cesarean section in our hospital. We compared the baseline demographics and clinical characteristics of patients with and without anemia. The correlation between prenatal anemia and obstetric blood transfusion-related parameters was evaluated using multivariate regression analysis.

RESULTS

A total of 749 patients were enrolled, with a mean prenatal hemoglobin level of 10.87 ± 1.37 g/dL. Among them, 54.87% (391/749) were diagnosed with anemia. The rate of obstetric blood transfusion was significantly higher in the anemia group (79.54%) compared to the normal group (44.41%). The median allogeneic red blood cell transfusion volume in the anemia group was 4.00 U (IQR 2.00-6.00), while in the normal group, it was 0.00 U (IQR 0.00-4.00). The prenatal hemoglobin levels had a non-linear relationship with intraoperative allogeneic blood transfusion rate, massive blood transfusion rate, red blood cell transfusion units, and fresh plasma transfusion volume in patients with placenta previa, with a threshold of 12 g/dL.

CONCLUSIONS

Our findings suggest that prenatal anemia is associated with a higher rate of blood transfusion-related parameters in women with placenta previa when the hemoglobin level is < 12 g/dL. These results highlight the importance of promoting prenatal care in placenta previa patients with a high requirement for blood transfusion.

Interactions between ultrasonographic cervical length and placenta accreta spectrum on severe postpartum hemorrhage in women with placenta previa

OBJECTIVE

To explore the interactions between cervical length (CL) and placenta accreta spectrum (PAS) on severe postpartum hemorrhage (SPPH) in patients with placenta previa.

METHODS

A retrospective case-control study was conducted at four medical centers in China, and 588 patients with placenta previa were included. The logistic regression analysis and restricted cubic splines (RCS) were used to evaluate the association between CL and SPPH. Furthermore, the joint effect of CL and PAS on SPPH was assessed, and the additive and multiplicative interactions were calculated.

RESULTS

After adjusting for potential confounders, the negative linear dose-response relationship was confirmed by RCS, and the change of odds ratio (OR) was more significant when CL was 2.5 cm or less. The risk of SPPH was significantly higher when CL of 2.5 cm or less co-existed with placenta increta/percreta than when CL of 2.5 cm less, or placenta increta/percreta existed alone (adjusted OR [aOR]_{CL ≤2.5cm&placenta accreta/non-PAS} 3.40, 95% confidence interval [CI] 1.37-8.45; aOR_{placenta increta/percreta&CL >2.5cm} 4.75, 95% CI 3.03-7.47; aOR_{CL ≤2.5cm&placenta increta/percreta} 14.51, 95% CI 6.08-34.64), and there might be additive interaction between CL and placenta increta/percreta on SPPH (attributable proportion due to interaction 50.7%, 95% CI 6.1%-95.3%).

CONCLUSION

If CL was routinely performed during PAS evaluation, the increased OR of short CL and PAS could allow better patient preparation through counseling.

Risk-factor model for postpartum hemorrhage after cesarean delivery: a retrospective study based on 3498 patients

This study aimed to investigate the risk factors of patients with postpartum hemorrhage (PPH) after cesarean delivery (CD) and to develop a risk-factor model for PPH after CD. Patients were selected from seven affiliated medical institutions of Chongqing Medical University from January 1st, 2015, to January 1st, 2020. Continuous and categorical variables were obtained from the hospital's electronic medical record systems. Independent risk factors were identified by univariate analysis, least absolute shrinkage and selection operator and logistic regression. Furthermore, logistic, extreme gradient boosting, random forest, classification and regression trees, as well as an artificial neural network, were used to build the risk-factor model. A total of 701 PPH cases after CD and 2797 cases of CD without PPH met the inclusion criteria. Univariate analysis screened 28 differential indices. Multi-variable analysis screened 10 risk factors, including placenta previa, gestational age, prothrombin time, thrombin time, fibrinogen, anemia before delivery, placenta accreta, uterine atony, placental abruption and pregnancy with uterine fibroids. Areas under the curve by random forest for the training and test sets were 0.957 and 0.893, respectively. The F1 scores in the random forest training and test sets were 0.708. In conclusion, the risk factors for PPH after CD were identified, and a relatively stable risk-factor model was built.

Development and internal validation of a nomogram to predict massive blood transfusions in neurosurgical operations

Objectives

A massive blood transfusion (MBT) is an unexpected event that may impact mortality. Neurosurgical operations are a major operation involving the vital structures and risk to bleeding. The aims of the present research were (1) to develop a nomogram to predict MBT and (2) to estimate the association between MBT and mortality in neurosurgical operations.

Material and Method

We conducted a retrospective cohort study including 3660 patients who had undergone neurosurgical operations. Univariate and multivariate logistic regression analyses were used to test the association between clinical factors, pre-operative hematological laboratories, and MBT. A nomogram was developed based on the independent predictors.

Results

The predictive model comprised five predictors as follows: Age group, traumatic brain injury, craniectomy operation, pre-operative hematocrit, and pre-operative international normalized ratio and the good calibration were observed in the predictive model. The concordance statistic index was 0.703. Therefore, the optimism-corrected c-index values of cross-validation and bootstrapping were 0.703 and 0.703, respectively.

Conclusion

MBT is an unexpectedly fatal event that should be considered for appropriate preparation blood components. Further, this nomogram can be implemented for allocation in limited-resource situations in the future.

Association of the placenta accreta spectrum score and estimated blood loss in placenta accreta spectrum patients with placenta previa: a retrospective cohort study

PURPOSE

The placenta accreta spectrum (PAS) score calculated by the scoring system may predict patients with PAS. We aim to find the relationship between estimated blood loss and the PAS score. Further, find the inflection point, identify PAS patients with placenta previa who were at risk for major bleeding.

METHODS

The PAS patients with placenta previa, as diagnosed by color Doppler ultrasound, were divided into two groups according to their PAS scores using a new scoring system. Blood loss, transfusion requirements, the rate of Intra-Abdominal Balloon Occlusion (IABO), and other indicators were analyzed between groups.

RESULTS

The estimated blood loss, intraoperative transfusion, postoperative transfusion, operation time, and hospitalization time significantly increased in the group with a PAS score ≥ 9 (P < 0.05). The inflection point analysis revealed that a significant increase in estimated blood loss occurred when the PAS score was beyond 10 (crude) or 6 (adjusted for age, body mass index, and IABO).

CONCLUSION

There was a non-linear relationship between estimated intraoperative blood loss and PAS score. When the PAS score was greater than 9, hemorrhage, the risk of major bleeding, the need for transfusions, and the placement of an abdominal aortic balloon all increase significantly.

Prevalence and Associated Factors for Post-Caesarean Delivery Blood Transfusion in Eastern Sudan: A Cross-Sectional Study

Background

Obstetric haemorrhage is a leading cause of maternal mortality and morbidity worldwide. Caesarean delivery (CD) is associated with significant blood loss, which may require blood transfusions. This study aimed to determine the prevalence and associated factors for post-CD transfusion.

Methods

A cross-sectional study was conducted in Gadarif maternity hospital, eastern Sudan, from March to September 2020. Sociodemographic, obstetric and clinical data, including pre- and postoperative haemoglobin levels, were collected. A multivariate logistic regression analysis was performed.

Results

A total of 539 women were enrolled in the study; the median (interquartile range) age of these women was 28.0 (8.0) years. The overall post-CD transfusion rate was 8.2%. Emergency CD (adjusted odds ratio [AOR]=2.57, 95% confidence interval [CI]=1.25‒5.28) and antepartum haemorrhage (AOR=44.70, 95% CI=11.18‒178.76) were associated with increased risk of post-CD blood transfusion. Preoperative haemoglobin (AOR=0.48, 95% CI=0.36‒0.64) and rural residence (AOR=0.45, 95% CI=0.22‒0.93) were associated with reduced risk for post-CD blood transfusion.

Conclusion

The overall prevalence of post-CD transfusion in this part of Sudan is 8.2%. Women with emergency CD, low preoperative haemoglobin levels and antepartum haemorrhage were at higher risk for post-CD transfusion. Risk identification and correction of antenatal anaemia can reduce the hazard of blood transfusion among CD women.

Prediction and Evaluation of Machine Learning Algorithm for Prediction of Blood Transfusion during Cesarean Section and Analysis of Risk Factors of Hypothermia during Anesthesia Recovery

Objective

To explore the application of machine learning algorithm in the prediction and evaluation of cesarean section, predicting the amount of blood transfusion during cesarean section and to analyze the risk factors of hypothermia during anesthesia recovery.

Methods

(1)Through the hospital electronic medical record of medical system, a total of 600 parturients who underwent cesarean section in our hospital from June 2019 to December 2020 were included. The maternal age, admission time, diagnosis, and other case data were recorded. The routine method of cesarean section was intraspinal anesthesia, and general anesthesia was only used for patients' strong demand, taboo, or failure of intraspinal anesthesia. According to the standard of intraoperative bleeding, the patients were divided into two groups: the obvious bleeding group (MH group, N = 154) and nonobvious hemorrhage group (NMH group, N = 446). The preoperative, intraoperative, and postoperative indexes of parturients in the two groups were analyzed and compared. Then, the risk factors of intraoperative bleeding were screened by logistic regression analysis with the occurrence of obvious bleeding as the dependent variable and the factors in the univariate analysis as independent variables. In order to further predict intraoperative blood transfusion, the standard cases of recesarean section and variables with possible clinical significance were included in the prediction model. Logistic regression, XGB, and ANN3 machine learning algorithms were used to construct the prediction model of intraoperative blood transfusion. The area under ROC curve (AUROC), accuracy, recall rate, and F1 value were calculated and compared. (2) According to whether hypothermia occurred in the anesthesia recovery room, the patients were divided into two groups: the hypothermia group (N = 244) and nonhypothermia group (N = 356). The incidence of hypothermia was calculated, and the relevant clinical data were collected. On the basis of consulting the literatures, the factors probably related to hypothermia were collected and analyzed by univariate statistical analysis, and the statistically significant factors were analyzed by multifactor logistic regression analysis to screen the independent risk factors of hypothermia in anesthetic convalescent patients.

Results

(1) First of all, we compared the basic data of the blood transfusion group and the nontransfusion group. The gestational age of the transfusion group was lower than that of the nontransfusion group, and the times of cesarean section and pregnancy in the transfusion group were higher than those of the non-transfusion group. Secondly, we compared the incidence of complications between the blood transfusion group and the nontransfusion group. The incidence of pregnancy complications was not significantly different between the two groups (P > 0.05). The incidence of premature rupture of membranes in the nontransfusion group was higher than that in the transfusion group (P < 0.05). There was no significant difference in the fetal umbilical cord around neck, amniotic fluid index, and fetal heart rate before operation in the blood transfusion group, but the thickness of uterine anterior wall and the levels of Hb, PT, FIB, and TT in the blood transfusion group were lower than those in the nontransfusion group, while the number of placenta previa and the levels of PLT and APTT in the blood transfusion group were higher than those in the nontransfusion group. The XGB prediction model finally got the 8 most important features, in the order of importance from high to low: preoperative Hb, operation time, anterior wall thickness of the lower segment of uterus, uterine weakness, preoperative fetal heart, placenta previa, ASA grade, and uterine contractile drugs. The higher the score, the greater the impact on the model. There was a linear correlation between the 8 features (including the correlation with the target blood transfusion). The indexes with strong correlation with blood transfusion included the placenta previa, ASA grade, operation time, uterine atony, and preoperative Hb. Placenta previa, ASA grade, operation time, and uterine atony were positively correlated with blood transfusion, while preoperative Hb was negatively correlated with blood transfusion. In order to further compare the prediction ability of the three machine learning methods, all the samples are randomly divided into two parts: the first 75% training set and the last 25% test set. Then, the three models are trained again on the training set, and at this time, the model does not come into contact with the samples in any test set. After the model training, the trained model was used to predict the test set, and the real blood transfusion status was compared with the predicted value, and the F1, accuracy, recall rate, and AUROC4 indicators were checked. In terms of training samples and test samples, the AUROC of XGB was higher than that of logistic regression, and the F1, accuracy, and recall rate of XGB of ANN were also slightly higher than those of logistic regression and ANN. Therefore, the performance of XGB algorithm is slightly better than that of logistic regression and ANN. (2) According to the univariate analysis of hypothermia during the recovery period of anesthesia, there were significant differences in ASA grade, mode of anesthesia, infusion volume, blood transfusion, and operation duration between the normal body temperature group and hypothermia group (P < 0.05). Logistic regression analysis showed that ASA grade, anesthesia mode, infusion volume, blood transfusion, and operation duration were all risk factors of hypothermia during anesthesia recovery.

Conclusion

In this study, three machine learning algorithms were used to analyze the large sample of clinical data and predict the results. It was found that five important predictive variables of blood transfusion during recesarean section were preoperative Hb, expected operation time, uterine weakness, placenta previa, and ASA grade. By comparing the three algorithms, the prediction effect of XGB may be more accurate than that of logistic regression and ANN. The model can provide accurate individual prediction for patients and has good prediction performance and has a good prospect of clinical application. Secondly, through the analysis of the risk factors of hypothermia during the recovery period of cesarean section, it is found that ASA grade, mode of anesthesia, amount of infusion, blood transfusion, and operation time are all risk factors of hypothermia during the recovery period of cesarean section. In line with this, the observation of this kind of patients should be strengthened during cesarean section.

The Feasibility and Safety of Temporary Transcatheter Balloon Occlusion of Bilateral Internal Iliac Arteries during Cesarean Section in a Hybrid Operating Room for Placenta Previa with a High Risk of Massive Hemorrhage

This study aimed to evaluate the feasibility and safety of temporary transcatheter balloon occlusion of bilateral internal iliac arteries (TBOIIA) during cesarean section in a hybrid operating room (OR) for placenta previa (PP) with a high risk of massive hemorrhage. This retrospective study analyzed the medical records of 62 patients experiencing PP with a high risk of massive hemorrhage (mean age, 36.2 years; age range 28-45 years) who delivered a baby via planned cesarean section with TBOIIA in a hybrid OR between May 2019 and July 2021. Operation time, estimated blood loss (EBL), amount of intra- and postoperative blood transfusion, perioperative hemoglobin level, hospital stay after operation, balloon time, fluoroscopy time, radiation dose, rate of uterine artery embolization (UAE) and hysterectomy, and complication-related TBOIIA were assessed. The mean operation time was 122 min, and EBL was 1290 mL. Nine out of sixty-two patients (14.5%) received a blood transfusion. The mean hemoglobin levels before surgery, immediately after surgery and within 1 week after surgery were 11.3 g/dL, 10.4 g/dL and 9.2 g/dL, respectively. In terms of radiation dose, the mean dose area product (DAP) and cumulative air kerma were 0.017 Gy/cm² and 0.023 Gy, respectively. Ten out of sixty-two patients (16.1%) underwent UAE postoperatively in the hybrid OR. One out of sixty-two patients had been diagnosed with placenta percreta with bladder invasion based on preoperative ultrasound, and thus underwent cesarean hysterectomy following TBOIIA and UAE. While intra-arterial balloon catheter placement for managing PP with a high risk of hemorrhage remains controversial, a planned cesarean section with TBOIIA in a hybrid OR is effective in eliminating the potential risk of intra-arterial balloon catheter displacement, thus reducing intraoperative blood loss, ensuring safe placental removal and conserving the uterus.

Developing and Validating Nomogram to Predict Severe Postpartum Hemorrhage in Women With Placenta Previa Undergoing Cesarean Delivery: A Multicenter Retrospective Case-Control Study

Objective

Developing and validating nomogram to predict severe postpartum hemorrhage (SPPH) in women with placenta previa (PP) undergoing cesarean delivery.

Methods

We conducted a multicenter retrospective case-control study in five hospitals. In this study, 865 patients from January, 2018 to June, 2020 were enrolled in the development cohort, and 307 patients from July, 2020 to June, 2021 were enrolled in the validation cohort. Independent risk factors for SPPH were obtained by using the multivariate logistic regression, and preoperative nomogram and intraoperative nomogram were developed, respectively. We compared the discrimination, calibration, and net benefit of the two nomograms in the development cohort and validation cohort. Then, we tested whether the intraoperative nomogram could be used before operation.

Results

There were 204 patients (23.58%) in development cohort and 80 patients (26.06%) in validation cohort experienced SPPH. In development cohort, the areas under the receiver operating characteristic (ROC) curve (AUC) of the preoperative nomogram and intraoperative nomogram were 0.831 (95% CI, 0.804, 0.855) and 0.880 (95% CI, 0.854, 0.905), respectively. In validation cohort, the AUC of the preoperative nomogram and intraoperative nomogram were 0.825 (95% CI, 0.772, 0.877) and 0.853 (95% CI, 0.808, 0.898), respectively. In the validation cohort, the AUC was 0.839 (95% CI, 0.789, 0.888) when the intraoperative nomogram was used before operation.

Conclusion

We developed the preoperative nomogram and intraoperative nomogram to predict the risk of SPPH in women with PP undergoing cesarean delivery. By comparing the discrimination, calibration, and net benefit of the two nomograms in the development cohort and validation cohort, we think that the intraoperative nomogram performed better. Moreover, application of the intraoperative nomogram before operation can still achieve good prediction effect, which can be improved if the severity of placenta accreta spectrum (PAS) can be accurately distinguished preoperatively. We expect to conduct further prospective external validation studies on the intraoperative nomogram to evaluate its application value.

Advancing Prediction of Risk of Intraoperative Massive Blood Transfusion in Liver Transplantation With Machine Learning Models. A Multicenter Retrospective Study

Background

Liver transplantation surgery is often accompanied by massive blood loss and massive transfusion (MT), while MT can cause many serious complications related to high mortality. Therefore, there is an urgent need for a model that can predict the demand for MT to reduce the waste of blood resources and improve the prognosis of patients.

Objective

To develop a model for predicting intraoperative massive blood transfusion in liver transplantation surgery based on machine learning algorithms.

Methods

A total of 1,239 patients who underwent liver transplantation surgery in three large grade lll-A general hospitals of China from March 2014 to November 2021 were included and analyzed. A total of 1193 cases were randomly divided into the training set (70%) and test set (30%), and 46 cases were prospectively collected as a validation set. The outcome of this study was an intraoperative massive blood transfusion. A total of 27 candidate risk factors were collected, and recursive feature elimination (RFE) was used to select key features based on the Categorical Boosting (CatBoost) model. A total of ten machine learning models were built, among which the three best performing models and the traditional logistic regression (LR) method were prospectively verified in the validation set. The Area Under the Receiver Operating Characteristic Curve (AUROC) was used for model performance evaluation. The Shapley additive explanation value was applied to explain the complex ensemble learning models.

Results

Fifteen key variables were screened out, including age, weight, hemoglobin, platelets, white blood cells count, activated partial thromboplastin time, prothrombin time, thrombin time, direct bilirubin, aspartate aminotransferase, total protein, albumin, globulin, creatinine, urea. Among all algorithms, the predictive performance of the CatBoost model (AUROC: 0.810) was the best. In the prospective validation cohort, LR performed far less well than other algorithms.

Conclusion

A prediction model for massive blood transfusion in liver transplantation surgery was successfully established based on the CatBoost algorithm, and a certain degree of generalization verification is carried out in the validation set. The model may be superior to the traditional LR model and other algorithms, and it can more accurately predict the risk of massive blood transfusions and guide clinical decision-making.

Effect of carboprost tromethamine injection combined with modified B-lynch suture and carboprost methylate suppositories in parturients with placenta previa

OBJECTIVE

To investigate the clinical value of carboprost tromethamine injection combined with modified B-lynch suture and carboprost methylate suppositories in the treatment of placenta previa parturients with postpartum hemorrhage after cesarean section.

METHODS

A total of 102 parturients with placenta previa and postpartum hemorrhage after cesarean section in our hospital were selected as the study subjects, and they were divided into Group A (carboprost tromethamine injection combined with modified B-lynch suture, n=35), Group B (carboprost methylate suppositories, n=34), and Group C (carboprost tromethamine injection, n=33) in accordance with a random number table. The amounts of hemorrhaging and clinical indices in the three groups were recorded, and the rescue effects were compared among the three groups.

RESULTS

The amount of hemorrhaging in Group A was significantly lower than that in Groups B and C during surgery and 24 h after surgery (P < 0.05). There were markedly improved clinical indices in Groups A, B and C, showing statistical significance (P < 0.05). There were statistically significant differences in hemostatic failure rate, hysterectomy, postoperative abdominal pain and puerperal infection between Groups A and B (P < 0.05). The intraoperative indices, postoperative infection, effective hemostasis rate and rate of advanced postpartum hemorrhage in Group A were remarkably higher than those in Groups B and C (P < 0.05), showing statistical significance (P < 0.05). There were statistically significant differences in blood oxygen saturation and pulse among the three groups before surgery and 2 h after surgery (P < 0.05).

CONCLUSION

Carboprost tromethamine injection combined with modified B-lynch suture and carboprost methylate suppositories can reduce the amount of hemorrhaging and the risk of postoperative infection in placenta previa patients with postpartum hemorrhage after cesarean section.

Comparison of Multivariable Logistic Regression and Other Machine Learning Algorithms for Prognostic Prediction Studies in Pregnancy Care: Systematic Review and Meta-Analysis

BACKGROUND

Predictions in pregnancy care are complex because of interactions among multiple factors. Hence, pregnancy outcomes are not easily predicted by a single predictor using only one algorithm or modeling method.

OBJECTIVE

This study aims to review and compare the predictive performances between logistic regression (LR) and other machine learning algorithms for developing or validating a multivariable prognostic prediction model for pregnancy care to inform clinicians' decision making.

METHODS

Research articles from MEDLINE, Scopus, Web of Science, and Google Scholar were reviewed following several guidelines for a prognostic prediction study, including a risk of bias (ROB) assessment. We report the results based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Studies were primarily framed as PICOTS (population, index, comparator, outcomes, timing, and setting): Population: men or women in procreative management, pregnant women, and fetuses or newborns; Index: multivariable prognostic prediction models using non-LR algorithms for risk classification to inform clinicians' decision making; Comparator: the models applying an LR; Outcomes: pregnancy-related outcomes of procreation or pregnancy outcomes for pregnant women and fetuses or newborns; Timing: pre-, inter-, and peripregnancy periods (predictors), at the pregnancy, delivery, and either puerperal or neonatal period (outcome), and either short- or long-term prognoses (time interval); and Setting: primary care or hospital. The results were synthesized by reporting study characteristics and ROBs and by random effects modeling of the difference of the logit area under the receiver operating characteristic curve of each non-LR model compared with the LR model for the same pregnancy outcomes. We also reported between-study heterogeneity by using τ2 and I2.

RESULTS

Of the 2093 records, we included 142 studies for the systematic review and 62 studies for a meta-analysis. Most prediction models used LR (92/142, 64.8%) and artificial neural networks (20/142, 14.1%) among non-LR algorithms. Only 16.9% (24/142) of studies had a low ROB. A total of 2 non-LR algorithms from low ROB studies significantly outperformed LR. The first algorithm was a random forest for preterm delivery (logit AUROC 2.51, 95% CI 1.49-3.53; I2=86%; τ2=0.77) and pre-eclampsia (logit AUROC 1.2, 95% CI 0.72-1.67; I2=75%; τ2=0.09). The second algorithm was gradient boosting for cesarean section (logit AUROC 2.26, 95% CI 1.39-3.13; I2=75%; τ2=0.43) and gestational diabetes (logit AUROC 1.03, 95% CI 0.69-1.37; I2=83%; τ2=0.07).

CONCLUSIONS

Prediction models with the best performances across studies were not necessarily those that used LR but also used random forest and gradient boosting that also performed well. We recommend a reanalysis of existing LR models for several pregnancy outcomes by comparing them with those algorithms that apply standard guidelines.

TRIAL REGISTRATION

PROSPERO (International Prospective Register of Systematic Reviews) CRD42019136106; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=136106.