A new approach for diagnosis of Acute Myeloid and Lymphoblastic Leukemia using gene expression profile and machine learning techniques

Leukemia is a cancer type in the world. One of the most accurate methods for detection and prediction of Acute Myeloid Leukemia and Acute Lymphoblastic Leukemia is to use DNA and genetic information of people. Microarray technology is a tool to study the expression of thousands of genes in shortest possible time. Analyzing the microarray datasets may not be possible without the statistical analysis and machine learning techniques. In this paper, microarray data sets and machine learning techniques are used for the diagnosis of leukemia.
The data used in this descriptive study are the expression of 7129 genes of 72 patients with leukemia which have been achieved by the microarray technology. Then, the diagnosis of AML and ALL was performed using the microarray data based on anisotropic radial basis function with the gain ratio and information gain.
The proposed method using information gain with the selection of 230 important genes and using gain ratio with the selection of 86 important genes was able to detect AML and ALL with accuracy of 97. 06%, whereas non-parametric kernel classification method based on the radial basis function has the accuracy of 35. 29٪ with 7129 genes.
The results of this study showed that the gene expression data and proposed method with gain ratio method are able to detect leukemia with high accuracy. Therefore, it seems that proposed method can help to accurately diagnose leukemia for a better decision making about the diagnosis of diseases and treatment of patients.