Anonymizing Prescription Data Against Individual Privacy Breach in Healthcare Database

Abstract

Prescription data is a subset of the health-related data which can be collected by drug store during the patient's medication period. In general, prescription data consists of a set of transaction records which contains patients name or patient’s identification number and their prescribed medicine name. Analyzing such data using data mining techniques brings various advantages for drug stores. However, performing data mining task is not trivial for the drug stores and possibly the drug store dispatches the prescription data to another party for data analysis. While it can solve the data analysis problem, unfortunately, such activity may result in privacy breach since sensitive information i.e., types of patients' disease due to the data miner has background knowledge to infer certain medicine to the disease type. To guarantee individual privacy protection while at the same time preserving database utility a method called data anonymization should be employed prior to handling the prescription data to another party for data mining purpose. A data anonymization which is based on swapping technique can be a solution to address the problem. Experimental results show that the swapping method successfully protects individual privacy with respect to reduce the number of item lost and maintain data utility of the anonymized database.