Machine Learning-Based Prediction Model of Preterm Birth Using Electronic Health Record

Figure 3

Classifiers used in this study. (1) Naive Bayesian (NBM): Naive Bayes calculates the posterior probability from , and ; is the posterior probability of class B and is the prior probability of predictor A and is the prior probability of class, and is the probability of the predictor for the particular class. (2) Support Vector Machine (SVM); SVM outputs a hyperplane that best separates the classes and has the largest separation of geometrical separations. (3) Logistic regression: The principle of logistic regression is to use a logistic function to map the results of linear regression between 0 and 1; is the input features, and is the weight of the features. is the predicted probability of class 1. (4) Artificial Neural Networks (ANN): An artificial neural network consists of an input layer, a hidden layer, and an output layer, and its core component is an artificial neuron. Each neuron is summed by several other neurons multiplied by weights; is the input features. (5) K-means: The K-Means algorithm minimizes the squared error for cluster ; is the unclassified sample, and is the clusters, and is the mean vector of clusters . (6) Random Forest Tree (RF): Random forest is an algorithm that integrates multiple decision trees through the Bagging idea of ensemble learning. The principle of random forest bagging is to vote the classification results of several weak classifiers to form a strong classifier.