Clustering of Drug Sampling Data to Determine Drug Distribution Patterns with K-Means Method : Study on Central Kalimantan Province, Indonesia

Background: Drug sampling and testing in the context of post-marketing control is an important component to ensure drug safety in the supply chains. The results are used by the Indonesian National Agency for Drug and Food Control (NA-FDC) for conducting public warnings, evaluating the Good Manufacturing Practice (GMP) and Good Distribution Practice (GDP) implementation, and enforcing the law against drug violation.Objective: This study aimed to identify and analyze drug distribution patterns to provide an overview of drug sampling in the public sector. Methods: The data was collected from Balai Besar Pengawas Obat dan Makanan (BBPOM) Palangka Raya’s database. The collected data were the drug sampling data from Integrated Information Reporting Systems (IIRS) application from 2014 to 2018. Next, we employed CRISP-DM methodology to analyze the data and to identify the pattern. K-means clustering model was selected for data modeling.Results: The dataset contained five attributes, i.e., drug name, therapeutic classes, district/city, sample category, and evaluation of drug surveillance. The drug distribution pattern formed three clusters. First cluster contained 522 drug items in eight therapeutic classes and spread over ten districts, second cluster contained 1542 drug items in five therapeutic classes and spread over five districts, and third cluster contained 503 drug items in eleven therapeutic classes and spread across nine districts.Conclusion: To conclude, the applied data mining technique has improved the decision on the drug sampling planning. It also provides in-depth information on the improvement of drug post-marketing control performance in Central Kalimantan Province.Keywords: Clustering, CRISP-DM, Data Mining, Drug distribution patterns, Drug quality control, Drug sampling