Privacy Preserving Data Mining Framework for Negative Association Rules: An Application to Healthcare Informatics

Protecting the privacy of healthcare information is an important part encouraging data custodians to give accurate records so that mining may proceed with confidence. The application association rule in has been widespread this point time. Most applications focus on positive rules, ignoring negative consequences particular diagnostic techniques. When it comes bridging divergent diseases and drugs, rules more helpful than ones. This especially true when physicians social organizations (e.g., a certain symptom will not arise symptoms exist). Data must be done way protects identity patients, dealing sensitive information. However, revealing puts at risk attack. Healthcare protection lately addressed by technologies disrupt (data sanitization) reconstruct aggregate distributions interest doing research mining. In study, metaheuristic-based sanitization for investigated order keep patient protected. It hoped using Tabu-genetic algorithm as optimization tool, suggested technique chooses item sets sanitized (modified) from transactions satisfy criteria goal minimizing changes original database. Experiments benchmark datasets show preserving (PPDM) method outperforms existing algorithms terms Hiding Failure (HF), Artificial Rule Generation (AR), Lost Rules (LR).