Identifying Disease Cluster from Blood Donor Data Using EDA and Graph Modelling
摘要
Blood transfusion is one of the key life-saving functions in comprehensive emergency health care. Obtaining safe blood for transfusions is the main challenge faced by the health workers. Delivering infected blood causes an increase in the mortality rate and the expense of patient care. The prompt availability of a compatible and safe blood group must be guaranteed. To achieve this, it is necessary to set up blood collection centres with screening facilities covering all geographies. At blood collection centres, screening tests for Transfusion Transmitted Infections (TTI) should be conducted after collecting donor information. The objective of the proposed study is to do a blood donor data analysis to identify disease clusters using Exploratory Data Analysis and Graph Modelling. The analysis is done on a real dataset of size 10,367 blood samples, extracted from the blood donation centre at Thiruvananthapuram, Kerala and its associated blood storage centres. The methodology for exploratory data analysis is done by framing a set of analysis points with the given dataset, and then analysing the distribution of attributes namely, TTI diseases, gender, age and pin code. The results are evaluated and validated with the TTI evidence from published literature. During the exploratory data analysis studies relationship between ABO with Rh-positive blood group and TTI, the distribution of ABO with Rh-group among donors and the distribution of donors by gender and age were identified. The feasibility of graph modelling techniques is analysed for disease cluster identification. Disease cluster analysis aids in identifying specific groups or populations at high risk for transfusion-transmitted infections. The information is very crucial for the targeted interventions and preventive measures.