Prediction of Anemia using Machine Learning Algorithms

Abstract

Anemia is a state of poor health where there is presence of low amount of red blood cell in blood stream. This research aims to design a model for prediction of Anemia in children under 5 years of age using Complete Blood Count reports. Data are collected from Kanti Children Hospital which consist of 700 data records. Then they are preprocessed, normalized, balanced and selected machine learning algorithms were applied. It is followed by verification, validation along with result analysis. Random Forest is the best performer which showed accuracy of 98.4%. Finally, Feature Selection as well as Ensemble Learning methods, Voting, Stacking, Bagging and Boosting were applied to improve the performance of algorithms. Selecting the best performer algorithm, stacking with other algorithms, bagging it, boosting it are very much crucial to improve accuracy despite of any time issue for prediction of anemia in children below 5 years of age.