Person Identification using Spatial Variation of Cardiac Signal

Data has become an absolute necessity due to the rapid developments in the field of Artificial Intelligence (AI) based systems. This requires user identification in almost all services, including healthcare services. ECG based biometry is an emerging technology that shows promising results as well as robustness against spoofing attack. Most of the works in this field have been done using one channel of ECG signal. Hence they don't use the information on spatial variation of ECG signal, which is unique to every person. In this work, we have proposed a long short term memory(LSTM) based multichannel data fusion technique that can exploit the spatial variation of the ECG signal. The results show that better accuracy can be achieved using less amount of enrollment data. We have achieved an accuracy of 98.77% and 99.29% for person identification in PTB and MIT-BIH arrhythmia database, respectively.