Multilabel Classification of Anemia Types Using Machine Learning: A Comparative Analysis of Problem Transformation Techniques and Classifiers.

Résumé fourni par la source

A blood count test (CBC) is said to be the primary and important test for a complete body health checkup, which plays a significant role in the primary diagnosis of bacterial (Neutrophils ↑) or viral (Lymphocyte ↑) infection, cancer, anemia, vitamin and mineral deficiency. Anemia is a sign of decreased blood flow in the body. The CBC parameter pattern is very complex so it is difficult to predict different types of anemia. In our research, a Multilabel classification model has been developed to predict anemia and its eight types IDA (Iron Deficiency Anemia), VitaminB12, Aplastic, SickleCell, FDA (Folate Deficiency Anemia), ACD (Anemia of Chronic Disease), Hemolytic, and Thalassemia. A Multilabel classification model has been developed using machine learning problem transformation’s three techniques: binary relevance, classifier chain, and label power set with six different classifiers (J48, Naïve Bayes, Logistic Regression, Random Forest, and SVM). Results provide interesting insights from classified data J48 algorithm in the classifier chain technique is proved to be the best compared to other methods.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.