Predictive Classification of ECG Parameters Using Association Rule Mining

Data mining is the procedure of extricating valuable information from the tremendous information stored in the database. Association rule mining is one of the most important and powerful data mining techniques. Association rule mining is normally carried out in two stages: first is to find frequent item set and second is to utilize those item sets to recognize the association rules. In recent medical history of cardiac arrests it has been observed that a huge gap exists in interpreting ECG data among differently skilled doctors. In this paper we will use the principle of meta-analysis and will reduce the gap between the interpretations of different doctors by employing statistical techniques like correlation and multiple linear regression. We would also generate rules using predictive apriori association rule mining among the various attributes of ECG to classify whether a patient requires an ECG before a cardiac arrest or not. The purpose of carrying out this work is to reduce the fatality rate and be able to predict that whether a patient requires ECG before actually facing a cardiac arrest and to minimize the cases of wrong interpretation.