A Machine learning-powered mobile application for linking blood donors to patients. a case of Mombasa, Kenya

Abstract

Access to blood for transfusions has always been a critical healthcare challenge, with statistic indicating that an estimated seven Kenyans require blood every ten minutes. However, less than half of these individuals receive transfusions when need, leading to a high number of deaths and complications that are preventable. This crisis is driven largely by inefficiencies in traditional methods of blood donor mobilization and recruitment. These conventional systems almost always result in delays, inadequate matching of donors to patients, and insufficient blood reserves during emergencies. As the demand for blood continues to rise, the need for a more efficient and responsive system has become essential. This research aimed to address the inefficiencies by developing a machine learning-powered mobile app that is tailored to specific healthcare needs of Mombasa County. The study employed an experiential design methodology, integrating descriptive and exploratory approaches to capture both qualitative and quantitative data, which ensured a thorough understanding of existing challenges and allowed for iterative development and testing. The application facilitates the linking of patients who require blood for transfusion to nearest potential blood donors, streamlining the donor-patient matching process and improving emergency response times. The findings demonstrate that the proposed system significantly reduces the time required to find compatible donors. Tests showed an average matching time of 15 seconds from request to donor identification, indicating a substantial improvement over traditional methods. The prediction accuracy of the model at 89.9% was satisfactory, indicating its potential for practical deployment. The system’s effectiveness was evaluated through testing, which confirmed its ability to accurately predict potential suitable donors and facilitate timely communication between patients and blood donors. This research contributes to the ongoing efforts to enhance blood donation systems by providing a novel approach that combines machine learning, geolocation, and geofencing technologies for improved donor-patient linkage. Keywords: Blood transfusion, machine learning, mobile application, experiential design, geolocation, geofencing, blood group.