<![CDATA[Over time and with the advancement of technology, an increasing number of disease-claim submissions have been received by Badan Penyelenggara Jaminan Sosial (BPJS) for Health, causing data accumulation to the point that the dataset can now be categorized as Big Data. One of the challenges of Big Data is that it cannot be processed using conventional methods, thus requiring specialized approaches such as data clustering. The purpose of this study is to determine the optimal number of clusters and to analyze the characteristics of the cluster groups. The type of data used is secondary data obtained from the BPJS Health database. The data used consists of claim data from Fasilitas Kesehatan Rujukan Tingkat Lanjutan (FKRTL) under BPJS Health from January 2019 to December 2020. The variables used include childbirth, accidents, catastrophic diseases, and other diseases. The stages of the clustering process include data normalization, parameter determination, application of the Density-Based Spatial Clustering of Application with Noise (DBSCAN) algorithm, and evaluation of cluster results using the silhouette index. The results of the clustering analysis on FKRTL claim data based on the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10), show that there are three clusters and one noise cluster, with an average silhouette index of 0.6595942, indicating that the model has a medium structure. Cluster 1 consists of two members with dominant claim categories being accidents and other diseases, cluster 2 consists of 27 members with childbirth as the dominant claim category, cluster 3 consists of four members with catastrophic diseases and other diseases as the dominant claim categories, and the noise cluster consists of one member with childbirth as the dominant claim category.]]>
Copyrights © 2025
