Predicting the postoperative blood coagulation state of children with congenital heart disease by machine learning
Kai Guo1, Xiaoyan Fu1, Huimin Zhang1
1Department of Transfusion Medicine, Beijing Children's Hospital, Capital Medical University, National Center for Children's Health, Beijing, China.
Insights
Machine learning accurately predicts postoperative blood coagulation in children with congenital heart disease (CHD), utilizing complete blood count (CBC) indicators. This approach offers a reliable alternative to conventional statistical methods for improved patient outcomes.
Area of Science:
- Medical Informatics
- Pediatric Cardiology
- Biostatistics
Background:
- Conventional statistical methods have been used for postoperative blood coagulation assessment in children with congenital heart disease (CHD).
- This study explores the application of machine learning (ML) for predicting postoperative blood coagulation function in pediatric CHD patients.
- An array of ML models were assessed to identify optimal prediction capabilities.
Purpose of the Study:
- To predict the postoperative blood coagulation function in children with congenital heart disease (CHD) using machine learning.
- To evaluate and compare the performance of various machine learning models for this prediction task.
- To identify key clinical indicators, particularly from complete blood count (CBC) tests, that are most influential in predicting postoperative coagulation status.
Main Methods:
- A retrospective data mining study involving 1,690 children with CHD.
- Screening of demographic characteristics, conventional coagulation tests (CCTs), and complete blood count (CBC) data.
- Application of machine learning algorithms including Decision Tree, Naive Bayes, Support Vector Machine (SVM), Adaptive Boost (AdaBoost), and Random Forest, with performance evaluated using ROC curves, calibration, and Lift curves.
Main Results:
- Machine learning models achieved significant predictive accuracy, with AUC values up to 0.82 for primary predictions.
- Improved models demonstrated true positive rates exceeding 80% in ROC curve analysis.
- The Support Vector Machine (SVM) model showed the best fit, with Lift values of 2.2 for abnormal and 1.8 for normal postoperative coagulation, indicating strong predictive power.
- Key predictors identified include age, sex, and CBC parameters such as MCV, MCH, MCHC, WBC, and PLT.
Conclusions:
- Machine learning and data mining algorithms show significant potential for predicting postoperative blood coagulation status in children with CHD.
- Complete blood count (CBC) indicators are particularly valuable for outcome prediction in real-world clinical data.
- ML-driven prediction offers a promising avenue for enhancing patient care and outcomes in pediatric CHD management.
Background:
Postoperative blood coagulation assessment of children with congenital heart disease (CHD) has been developed using a conventional statistical approach. In this study, the machine learning (ML) was used to predict postoperative blood coagulation function of children with CHD, and assess an array of ML models.
Methods:
This was a retrospective and data mining study. Based on the samples of 1,690 children with CHD, and screening data based on demographic characteristics, conventional coagulation tests (CCTs) and complete blood count (CBC), with a precise data selection process, and the support of data mining and ML algorithms including Decision tree, Naive Bayes, Support Vector Machine (SVM), Adaptive Boost (AdaBoost) and Random Forest model, and explored the best prediction models of postoperative blood coagulation function for children with CHD by models performance measured in the area under the receiver operating characteristic (ROC) curve (AUC), calibration or Lift curves, and further verified the reliability of the models with statistical tests.
Results:
In primary objective prediction, as decision tree, Naive Bayes, SVM, the AUC of our prediction algorithm was 0.81, 0.82, 0.82, respectively. The accuracy rate of the overall forecast has reached more than 75%. Subsequently, we furtherly build improved models. Among them, the true positive rate of the AdaBoost, Random Forest and SVM prediction models reached more than 80% in the ROC curve. These overall accuracy rate indicated a good classification model. Combined calibration curves and Lift curves, the better fit is the SVM model, which predicted postoperative abnormal coagulation, Lift =2.2, postoperative normal coagulation, Lift =1.8. The statistical results furtherly proved the reliability of ML models. The age, sex, mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), white blood cell count (WBC) and platelet count (PLT) were the key features for predicting the postoperative blood coagulation state of children with CHD.
Conclusions:
ML technology and data mining algorithms may be used for outcome prediction in children with CHD for postoperative blood coagulation state based on the bulk of clinical data, especially CBC indictors from the real world.
