Predicting Bone Marrow Transplantation Outcomes Using Decision Trees: A Supervised Learning Approach

Bone marrow transplantation (BMT) is a critical therapeutic intervention for patients with hematological malignancies and other life-threatening disorders. Accurately predicting the outcomes of BMT is essential for improving patient care, optimizing treatment strategies, and managing post-transplantation risks. In this study, we present a supervised learning approach using decision trees to predict the outcomes of bone marrow transplantation based on clinical and demographic data. We collected data from patients who underwent bone marrow transplantation, including variables such as age, gender, disease type, donor type, pre-transplant health indicators, and post-transplant complications. Decision trees, a widely used machine learning algorithm, were employed to build predictive models. These models aim to identify key factors that influence transplantation success, such as survival rate, disease relapse, graft-versus-host disease (GVHD) occurrence, and long-term patient health outcomes. Our results demonstrate that decision trees can effectively classify patients into outcome categories with reasonable accuracy. Important predictors identified include patient age, donor compatibility, and specific pre-transplant health conditions. By providing interpretable decision paths, the model aids clinicians in understanding the risk factors associated with poor or favorable outcomes. Moreover, the decision tree approach allows for personalized predictions, supporting more tailored patient care.