To search, Click below search items.

 

All Published Papers Search Service

Title

Automated Diagnosis of Iron Deficiency Anemia and Thalassemia by Data Mining Techniques

Author

Maysam Hasani, Ali Hanani

Citation

Vol. 17  No. 4  pp. 326-331

Abstract

In the present paper, three types of anemia including iron deficiency anemia (IDA), β-thalassemia trait and α-thalassemia trait (cis and trans) have been investigated. Detection of these three types of anemia is difficult, as their blood characteristics are similar to each other. Also, specialists use some tests to diagnose these disorders that those tests are very time consuming and costly. Thus, providing a model for accurate diagnosing of these kinds of anemia is extremely important. The present study mainly focused on a simple complete blood count (CBC) test instead of using some tests to detect and differentiate between these kinds of anemia in Weka software. In order to suggest an algorithm with the highest accuracy and the lowest mean absolute error, five classification algorithms and a vote algorithm were used and the performance of the vote algorithm was compared with the performance of those five algorithms. The results of this study indicated that combining J48, IBK and Naive Bayes algorithms using voting algorithm with all the features and reduced features had the highest performance with the accuracy of 96.343 and 96.2169, respectively. Using hybrid algorithm (vote) demonstrated that hybrid algorithm increases diagnosis accuracy and decreases error rate in comparison with the single classifiers.

Keywords

α-thalassemia trait, β-thalassemia trait, Iron deficiency anemia (IDA), Vote algorithm, Complete blood count (CBC), Data mining techniques

URL

http://paper.ijcsns.org/07_book/201704/20170444.pdf