Abstract

This paper proposes the general framework for mining sequences from health care database. The database is a relational model consisting of set of temporal records of individual patient consisting of basic information of the patient ie Patient_ID, age, gender etc. the second part is a series of sequences representing the set of treatment given to the patient during regular visit to the doctor and the third part is class label. Similarity search of sequences is performed to convert the database of sequences, to the database of items, so that apriori algorithm can be applied. Weighted association rule mining has been performed to find the frequent sequence of treatment provided to the patient. Classification association rules (CAR) having positive class label as consequent, represents the frequent sequence of treatment given to the patient for successful treatment. With the experimental results, author feels confident in declaring that the framework is feasible in the medical domain.