Data preparation services for the authentication of people by their electrocardiogram

Abstract

For modern telemedicine, it is important to solve the problem of patient authentication. This paper studies the possibility of using personal unique features of people, namely their electro-cardiogram (ECG), for authentication. Biometric authentication is promising in telemedicine, where patients send physicians through information channels increasing amounts of their data. When some clinic monitors its patient remotely and asks them for an ECG, it should be confi-dent that the received ECG belongs to that particular patient. Biometric authentication using ECG is a rather specific area of research and now there are no leading technologies in this field. This method is the goal of the technology being developed. In this direction, we have already conducted research, developed algorithms, and tested their viability on test data. Combining the obtained algorithms and machine learning to solve the problem of classification of ECG belong-ing to a person, there was developed a program capable of authenticating people by their ECG. This gave confidence in the possibility of using such technology. The next step was to scale this technology to reach two goals: to test the performance of the technology on a much larger amount of data and to prepare it for use in real conditions when processing time should be min-imal. The paper describes the gradual development from a program running on a personal com-puter to a system performing computing in a cloud environment. The time spent on conducting experiments and comparing the running time of programs on a personal computer and in a cloud environment are shown. The most efficient configuration of the system for cloud architecture was found experimentally.