We appreciate the opportunity to respond to the three Letters to the Editor [1-3] commenting on our article [4], published in the Journal of Internal Medicine. We have carefully discussed them with the respective authors. We want to express our sincere gratitude for their interest in our work and for the valuable suggestions they have provided. First, we would like to address the concern raised in all three letters regarding the absence of clear exclusion criteria for comorbidities in our study group [4, 5]. We fully acknowledge and agree with the points raised by the authors of the letters. Although we share their concerns, our study was based on a retrospective data analysis. Consequently, we opted to address these effects by applying statistical corrections in a multivariate analysis rather than excluding them from our study altogether. We agree that in future prospective studies on this topic, it is indeed crucial to thoroughly consider and incorporate multiple factors that may influence patient outcomes into the exclusion criteria. Second, we would like to address their comments regarding the significant prehospital characteristics of out-of-hospital cardiac arrest (OHCA) patients highlighted in the study. We concur with the authors that there are numerous factors specific to OHCA patients, as opposed to in-hospital cardiac arrest (IHCA) patients, that can influence outcomes [6]. Indeed, it may be more effective to analyse OHCA patients separately from IHCA patients to reach more definitive conclusions. OHCA patients often face more challenges that can delay extracorporeal membrane oxygenation (ECMO) initiation, and in our data, these factors were associated with a poorer prognosis compared to IHCA patients. In future research, it would be advantageous to analyse OHCA patients separately from IHCA patients using a larger dataset to derive clearer insights on these issues. Third, we would like to comment on the issues raised by the authors regarding the specific details of post-cardiac arrest care. We agree that various post-cardiac arrest interventions—such as blood transfusions, ventilator settings, treatment of infectious complications, and therapeutic temperature management—as well as complications occurring during ECMO, such as insertion site bleeding, limb ischemia, and intracranial hemorrhage, are critical factors that impact the prognosis of cardiac arrest patients [7, 8]. Unfortunately, our study lacked sufficient data in this area to present detailed results. We recognize that detailed descriptions of these post-cardiac arrest treatments and complications may be crucial in understanding the prognosis of ECPR patients, and future studies should include these details and better assess their impact on outcomes. Finally, we would like to respond to the point raised by the authors regarding the insufficient evaluation of long-term outcomes beyond 6 months [6]. To include as many ECPR patients as possible in our study, we analysed data from patients who had recently undergone ECPR, resulting in a relatively short observation period of about 6 months. We agree that this short observation period may have limited our ability to assess long-term outcomes beyond 6 months. In future studies, we believe that it would be highly valuable to analyse patient outcomes over a longer follow-up period of 1 year or more. Despite the many limitations pointed out by the authors, we believe our study is significant in highlighting the importance of timely ECMO intervention in patients with cardiac arrest. We are confident that as more data on ECPR become available, higher quality analyses will help clarify some of the unknowns. Finally, we would like to express our deepest gratitude to the reviewers for their interest in our research and their valuable advice. Sang-Wook Lee: Writing—original draft; conceptualization; writing—review and editing; investigation. Ji-Hoon Sim: Writing—review and editing. The authors declare no conflicts of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Despite recent advances in surgical techniques and perinatal management in obstetrics for reducing intraoperative bleeding, blood transfusion may occur during a cesarean section (CS). This study aims to identify machine learning models with an optimal diagnostic performance for intraoperative transfusion prediction in parturients undergoing a CS. Additionally, to address model performance degradation due to data imbalance, this study further investigated the variation in predictive model performance depending on the ratio of event to non-event data (1:1, 1:2, 1:3, and 1:4 model datasets and raw data).The area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC) were evaluated to compare the predictive accuracy of different machine learning algorithms, including XGBoost, K-nearest neighbor, decision tree, support vector machine, multilayer perceptron, logistic regression, random forest, and deep neural network. We compared the predictive performance of eight prediction algorithms that were applied to five types of datasets. The intraoperative transfusion in maternal CS was 7.2% (1020/14,254). XGBoost showed the highest AUROC (0.8257) and AUPRC (0.4825) among the models. The most significant predictors for transfusion in maternal CS as per machine learning models were placenta previa totalis, haemoglobin, placenta previa partialis, and platelets. In all eight prediction algorithms, the change in predictive performance based on the AUROC and AUPRC according to the resampling ratio was insignificant. The XGBoost algorithm exhibited optimal performance for predicting intraoperative transfusion. Data balancing techniques employed to alter the event data composition ratio of the training data failed to improve the performance of the prediction model.