Data mining to improve safety of blood donation process

Abstract

Blood donation is a voluntary activity in the United States and provides critical blood units for transfusions. Blood is collected, processed and transported by blood centers to hospitals, though some hospitals also collect blood directly from donors. Blood donation is very safe, but a small percentage of donors can have reactions and some of these reactions can lead to serious injury. Donor hemovigilance is the surveillance and analysis of donor reactions with the goal of understanding the factors influencing reactions and indentifying steps to improve donor safety [9]. Historically in the U.S., donor hemovigilance is managed by individual collection centers to improve their specific organization's donation processes. The Department of Health and Human Services (HHS) has developed a software tool (called DonorHART™) to collect, organize and analyze reactions that occur at different participating blood centers and hospitals [6, 8]. Data mining is used to analyze factors influencing the donor reactions, and the insights are shared with the community to help blood center and hospital managers and quality improvement administrators undertake interventions to improve donor safety [10]. The paper presents the history of the donor hemovigilance development and two data mining efforts performed on the data collected to improve the safety of blood donation processes. © 2014 IEEE.