Prediction of Pregnancy Complications Using Machine Learning and Deep Learning Algorithms

Abstract

Machine learning algorithms can assist pregnant women and physicians in predicting risk factors associated with pregnancy. In order to reduce maternal mortality, early and accurate detection of pregnancy related complications is essential. In this chapter, a systematic evaluation of prior work based on pregnancy risk prediction using deep learning and ML is carried out. Based on this the work proposes models prepared using three machine machine learning algorithms: SVM, DT and extra tree classifier, and two deep learning algorithms: LSTM and Bi-LSTM. All the models were trained to identify the pregnancy related risk in three categories: low, medium and high. The proposed method is implemented in three different steps. First step includes of data selection and pre-processing, in second step implementation of ML and DL algorithms and in third step models were evaluated to test the performance using standard metrics. The results indicate Bi-LSTM model outperformed by achieving 94.34% of accuracy compared to other models.