Design of a Fingertip-Based Electrocardiography System as a Biometric Identifier Using an Online KNN Classifier

Acquiring electrocardiography (ECG) signals as biometric tools can be challenging in various aspects. Nowadays, biometric systems are expanding and replacing traditional methods. In this study, a fingertip-based electrocardiography biometric identification method combined with biofeedback is proposed, in which data were collected from six healthy young females aged ۲۵ ± ۵ years and four healthy males, including two subjects aged ۴۰ ± ۵ years and two subjects aged ۲۰ ± ۵ years. After performing preprocessing steps including frequency filters, time-domain features of the signal were extracted and fed into a KNN classifier. This classifier classified the received data with ۱۰۰% accuracy. In addition, after classification, this device was able to transmit messages to a computer through a serial communication system. Furthermore, it appears that this device can be provided to the public as a low-cost, accessible, and easy-to-use biometric tool. This device can send various commands depending on the cardiac status of individuals. In future studies, this system can be implemented on reinforcement learning and unsupervised classifiers so that the systems do not require the provision of initial labels.

Keywords:

Biometrics , KNN , Electrocardiography , Biofeedback , Arduino