Prediction of Hypertensive Disorders in Pregnancy using Machine Learning Algorithms

Objective: The primary objective of this study was to evaluate the efficacy of the HDP gestosis score by adopting Machine Learning Algorithms (MLA) to predict preeclampsia. Methods: The information was gathered retrospectively by reviewing the patient’s records. The HDP Gestosis Score predictors were evaluated for their ability to predict preeclampsia. The following machine learning models were utilized in this study: XG Boost classifier, Naive Bayes classifier, Support Vector Machine (SVM), RF classifier, and Decision Tree to evaluate the predictive efficiency of the HDP factors in predicting hypertensive diseases during pregnancy. Results: Approximately 55% of women were assessed as being at high risk for preeclampsia, 17% at low risk, and the rest at moderate risk. Compared to other models, Naive Bayes (NB) and Random forest classifier performed better for prediction of women at risk for preeclampsia. The precision and accuracy of NB were 0.97 and 0.96, respectively, with an AUC of 0.96. The precision and specificity of RF were 0.88 and 1.0; correspondingly, with an AUC of 0.82, both models demonstrated greater under the ROC curve values. Conclusion: Gestosis score and the MLA are effective models for predicting preeclampsia.