A Novel Approach for Finding Frequent Medicine Set Using Maximal Apriori for Medical Application - Proceedings of All India Seminar on Biomedical Engineering 2012 (AISOBE-2012) - page 135

Introduction
This chapter emphasizes the uniqueness of medical data mining. Raw medical data
are voluminous and heterogeneous. Medical data may be collected from various
images, interviews with the patient, laboratory data, the physician's observations
and interpretations, and physician's recommendations for disease. All these
components may bear upon the diagnosis, prognosis, and treatment of the patient,
and cannot be ignored. All recommendations of medicine given by the physician
can be used to analyze frequently used medicine (Frequent Medicine Set). This
information can be used by the physician, Medical Store's owner, and pharmacy
companies for decision making, regarding human health. Here, we are trying to
implement SRMine Algorithm to find ''Frequent Medicine Set''.
This chapter includes three sections: first, SRMine Algorithm, second, example,
and lastly conclusion.
SRMine Algorithm
Input: Transactional Database D, minimum support count [ 1 ].
Step 1: The algorithm scans the database in order to count the number of
occurrences of each item to find the candidate 1-itemset with their sup-
port count.
Step 2: The set of frequent 1-itemset L1 can then be determined by removing the
items having less than the minimum support count. It consists of the
candidate 1-itemsets satisfying minimum support. Let the number of
frequent 1-itemset be ''n''.
Step 3: Removes the infrequent items from each transaction and counts the
number of items in each transaction (item_count).
Step 4: The transactions are sorted in descending order based on the item count.
Step 5: Call SRMine(Database).
Step 6: Stop.
(A) SRMine(Database) Procedure
Step 1: Take 2-D array; put the transaction into 2-D array with their count of
repetition.
Step 2: Find maximal transactions (k-itemset) from the array whose count is
greater than or equal to the minimum support known as maximal frequent
itemsets or transactions. If k-itemsets count is less than minimum support,
then look for k-itemsets and (k - 1)-itemsets jointly for next (k - 1)
maximal itemsets and so on until no itemsets count is found greater than
minimum support.