Determination of anemia and anemia related factors in pregnant women with supervised machine learning methods

2. ABSTRACT Determination Of Anemia And Anemia Related Factors İn Pregnant Women With Supervised Machine Learning Methods The severe level of anemia during pregnancy requires the determination of sociodemographic characteristics affecting anemia and combating these factors. While detecting these factors in recent years, machine learning and classification algorithms provide great benefits to users. In this thesis, it is aimed to determine anemia and factors related to anemia in pregnant women by using supervised machine learning methods and to establish a clinical decision support system in this regard. In order to reflect all pregnant women living in Elazig city center, approximately 489 pregnant women were randomly selected from 3228 general pregnant population registered in family health centers according to population size. Using the Weka software from the data obtained as a result of the study, classification models that can detect anemia with supervised machine learning methods were created. Anemia detection system was created with rule-based Jrip, OneR and PART algorithms. The success metrics of these algorithms are accuracy, Kappa statistic, mean absolute error, root mean squared error, relative absolute error, root relative squared error, TP rate, FP rate, precision, recall, F-measure, MCC, ROC area and PRC area obtained according to the method. In the first method, all of the data was used for training, while in the second method, 5-fold cross-validation method was used, and in the third and last method, 70% of the data was obtained as a training set. Anemia was detected with accuracy values of 96.36% for Jrip algorithm, 85.45% for OneR algorithm and 97.98% for PART algorithm used in the study. Among the variables used, tea consumption and dark tea preference, number of pregnancies, advanced maternal age and iron supplementation were found to be the most important risk factors associated with anemia. Keywords: Machine Learning, Classification, Anemia