Data Mining Hospital Treatment and Discharge Summary of Sickle Cell Disease Patients

Abstract

Sickle cell disease (SCD) is a hereditary blood disorder that affects certain parts of the world. This disease affects hemoglobin, causing red blood cells to change shape, such as sickle and crescent, making it difficult to supply oxygen to all of the human body's cells. Various genotypes of SCD have been discovered; the most common disorder is sickle cell anemia. This study is a continuation of our ongoing research on 191,406 clinical records of SCD patients who visited and got hospitalized over a 12-year period (between 2008 - 2020). This paper focused on conducting the retrospective analysis and then applying data mining classification algorithms on SCD patients' data based on hospitalization records, hospital visits, hospital admissions reasons, department patients were admitted to, the length of time patients were treated in the hospital, blood transfer section for S C D patients, and discharge reason for different types of S C D patients. Five distinct classification models with ten cross-validations were experimented using the Naive Bayes, J48, SVM, NN, and PART algorithms. Furthermore, parameter optimization was carried out to determine the optimal classification results of each algorithm. Naïve Bayes with an accuracy of 95.50%, was faster, correctly classified clinical cases, and provided detailed correlation results for each of the target features. Finally, we extracted knowledge clusters on hospital clinical services for SCD patients, which were then validated by medical doctors in order to better serve SCD patients visiting the hospital.