A clustering based system for instant detection of cardiac abnormalities from compressed ECG

► In this paper, we have reported an innovative mechanism of cardiovascular disease detection directly from the compressed ECG. ► We have experimented on 20 different ECG segments, where 13 ECG segments were normal and 7 ECG segments were abnormal. ► The proposed data mining based approach could classify all the 13 normal and 7 abnormal ECG signals directly from the compressed ECG (i.e. 100% accuracy was obtained without performing decompression). ► Our proposed algorithm also successfully identified the initialization of abnormality, where in a particular segment normal and abnormal ECG signal was present.