Efficient clustering technique for feature selection in cancer gene data

Abstract

In the health care sector, medical and biomedical industry is focused on developing innovative tools to support medical practitioners. Among the various developments being carried out in the field of medical and biotechnology, disease prediction is an emerging and needful area where the medical diagnosis supportive system has to process the huge amount of genomic input dataset to identify the type of disease and classify. It is important because the early predictions of some tumors can be curable and reduces the death rate due to late suspect of the disease. For example, if the altered sequence of DNA of tumor infected person is accurately identified and can predict the next mutation of the disease, it is possible to rearrange the DNA sequence and cancer can be treatable genetically. Analysis of Genome sequences not only discloses the brief understanding of the evolution of disease and its related mechanisms but also can acts as a primary factor for development of new drugs in the near future. Due to the large size of genome sequence, machine learning plays a crucial role in analysis of the data and eventually in prediction of the disease. Among machine learning technique Clustering techniques have been effectively used not only for identifying the optimal genes for classifying a specific disease but also have been used for efficient prediction. The challenges in using microarray data include the availability of small samples with respect to number of genes which drastically impacts the performance of the underlying machine learning algorithms. To address this issue selecting the optimal gene is essential to improve the classification accuracy. newline