Classification of data obtained from portable ECG devices using support vector machines

We have developed an algorithm to classify Electro-Cardio-Gram (ECG) data as Arrhythmic or Normal. The ECG graph obtained is processed and the positions of P, Q, R, S, T peaks, lengths of RR interval, P-wave, QRS complex, ST segment and T-wave in each beat are extracted. The extracted features are used for training the Support Vector Machines (SVM) which could be later employed for classifying unknown data. This method can be further extended to diagnose heart diseases. We anticipate that the classification algorithm when implemented in the portable ECG device can be extremely useful for preliminary diagnosis of heart diseases in rural areas where access to qualified medical professionals is limited.