T.Ragunthar1, Dr.S.Selvakumar2
1 Dept. of Computer Science and Engineering/Sri Sai Ram Institute of Technology, Chennai,India
Email:ragunthar@gmail.com
2 Dept. of Computer Science and Engineering /GKM College of Engineering and Technolgy, Chennai,India
Email:sselvakumar@yahoo.com
Abstract - Bone Marrow Transplant(BMT) tends to make the patients susceptible to various risk factors, which was obtained from the analysis of the various pre and post transplants, in turn it enhanced the better understanding of the factors influencing success. Collaborative techniques were used to collect highly sparse records with missing values per transplant. The performance of various classification algorithms trained on predicting the survival status was evaluated using Ten-Fold-Cross validation. High accuracy levels were prominently shown for the operations that had the highest chances of success. Direct application in the prioritization of resources and in-donor matching was identified for the patients with highest chances of survival. For patients where …show more content…
The underlying diseases leading to the transplant include thalassemia and various forms of leukemia, categorized into five classes of acute lymphoblastic leukemia (ALL), acute myeloge nous leukemia (AML), chronic myelogenous leukemia (CML), plasma cell leukemia (PCL), or others.
The pretransplant attributes consist of categorical variables, ordinal variables, and calendar dates. The named categorical variables are derived from records and measurements and include the patient’s gender and blood type. The ordinal values include, for example, the level of various antigens, such as the Human Leukocyte Antigens, in the patient, the donor, and the patient’s parents. The date of birth and that of the diagnosis and the transplant are also