A Comparative Study for Prediction of Hematopoietic Stem Cell Transplantation-Related Mortality

Multipotent hematopoietic stem cells are transplanted during hematopoietic stem cell therapy (HSCT) in order to replicate inside of a patient and produce more healthy blood cells. These cells are often taken from bone marrow, peripheral blood, or umbilical cord blood. Patients with particular blood or bone marrow malignancies, like multiple myeloma or leukemia, are those who have the procedure the most frequently. Though a lifesaving procedure, it comes with its risk; hence, to minimize the risk, prediction of survivability and the factors affecting it is crucial. In this study, the authors have done an extensive and rigorous analysis of various works done in predicting the mortality of patients undergoing hematopoietic stem cell transplantation that involves the use of machine learning and data mining techniques and have compared them. This study gives an overview of the available machine learning and data mining techniques in improving risk prediction for patients undergoing HSCT that provides an alternative to traditional risk scores and indexes. Though these techniques already outperform the in-use risk scores and indexes, they can be still improved upon by the use of a specialized and large amount of data.