PREDICTION OF COMPLICATION OCCURANCE IN PREGNANCY USING MACHINE LEARNING

This research study aims to develop a machine learning model that can predict various pregnancy complication occurrence that affect women (pregnant women), driving to show the performance and accuracy level of each machine learning algorithm implemented. The techniques used in this research are the Support Vector Machine (SVM), Random Forest (RF) and the Artificial Neural Networks (ANN) with the data sourced from Kaggle repository. There were 3525 instances with 17 features in the case of gestational diabetes mellitus and 239 instances with 27 features in the case of pre-eclampsia in the datasets. A comparative analysis was first done to check the predictive Accuracy level of each model alongside its Recall, precision and F1-Score where Random forest came out to have performed the best with accuracy score of 0.96 and 0.85 for both complications respectively. To further provide a more robust and accurate model, in line with the objective of this research a hybrid model was then built merging the effectiveness of the Random forest and artificial neural network using various hybrid techniques. An increase in the accuracy score was shown both in gestational diabetes mellitus and especially in the case of Pre-eclampsia. At the end of this research it was discovered that Random forest performed a great role in the prediction of these complications individually and the hybrid model outshined in its own performance, choosing the hybrid model to be used in the implementation of the model.