Abstracto

Post Market Drug Analysis using Irregular Pattern Mining Scheme

Mr. S. Prakash, Ms. S. Kanjanadevi

Frequent pattern mining is performed using association rule mining algorithms. Candidate-set and item-set are prepared using the attribute name and its associated values. Minimum support and confidence values are used to select frequent patterns. Frequent pattern mining methods produces better performance in sparse or low dimensional data values. Dense and high-dimensional data sets have to use high thresholds to produce results within limited time and low support patterns. Rule mining methods are used to identify the drug reactions on patients. Drug reaction analysis is performed to find out the casual associations between two set in low frequency levels. Knowledge-based approach is used to capture the degree of causality of an event pair within each sequence with application or domain dependent. Interestingness measure incorporates the causalities across all the sequences in a database. Premarketing analysis is not sufficient to detect rare areas. A data mining framework is used to mine causal associations in patient data sets where the drug-related events of interest occur infrequently. A computational fuzzy Recognition-primed Decision (RPD) model is used to estimate the interestingness measure. Support count estimation algorithm is used to estimate support count for each drug. Pair generation algorithm is used to prepare candidate set pairs. Class leverage detection algorithm is applied to mine the causal relationship between drugs and their associated adverse drug reactions (ADRs). Scalability feature is provided in the enhanced drug reaction analysis system. Class leverage detection algorithm is enhanced with SQL functions. Rule summary analysis mechanism is integrated with the system to improve the accuracy levels. Support estimation and candidate pair generation algorithm are enhanced with aggregation functions.