Discovering drugs combination pattern using FP-growth algorithm

A drug can be used to deal more than one diseases and to deal an illness often need a combination of more than one drugs. This paper present how to discover a pattern of a combination of medicines related to a diagnosis of diseases using FP-Growth one of frequent pattern mining algorithm. We use FP-Growth because it has better performance than Apriori and Eclat. Data is collected from outpatients pharmacy of Sukoharjo state hospital, Central Java, Indonesia during January 2015 to June 2016 and obtain 526,195 records of prescription data and use a diagnosis of diseases base on the ICD-10 standard. This studies just apply on the top ten of the most frequently occurred illness in the outpatient's services of Sukoharjo state hospital. Then the pattern of association between diseases and combination of drugs was reviewed by pharmacist committed to being validated. These studies result in some combination of medicines for to top ten of the most frequent diseases. We also found 21 similar combinations of drugs for various diseases. In the future, this finding can be used to provide suggestions to physicians to select an appropriate mix of the drug to deal some diseases.