Managing voluntary interruption of pregnancy using data mining

Abstract

When a woman aims to terminate an unplanned pregnancy, she must go to a specialized healthcare unit, such as Júlio Dinis Maternity Hospital. In this unit, the procedures of voluntary interruption of pregnancy are done by two kinds of drug administration: the first one is always done by a nursing team, the second one can be performed at home or by a nursing team, depending on patient features. It is important to give the best option to the pregnant. In this paper, it is proposed to predict whether the second drug phase is done at home or at the hospital. The use of Data Mining (DM) helps in performing this step. Throughout this study, DM models capable to make predictions in a real environment using real data were induced. It was adopted the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology. Four distinct techniques were considered: Decision Tree (DT), Naïve Bayes (NB), Support Vector Machine (SVM) and Generalized Linear Models (GLM) to perform classification tasks. Using these techniques it was possible to obtain acceptable results for each model. A value greater than 89% of accuracy and 91% in sensibility was achieved in some models