The idea behind this rule mining algorithm is finding those rules that satisfy certain requirements, such as the minimum support in the dataset and the minimum confidence, rather than finding all the rules from a database and then pruning them [Li et al. 2001]. Chen et al. [2005]introduced two modifications in order to find patterns of a minority class, which was the patients with the potential to experience an ADR, rather than a majority class,which would not experience the ADR for a particular drug. To achieve this, two factors were introduced: local support and risk ratio.