An Echo State Network-Based Method for Identity Recognition with Continuous Blood Pressure Data

With the development of Continuous Blood Pressure (CBP) monitoring devices, we can collect real-time blood pressure non-invasively and accurately. Since CBP data can reflect the unique dynamical characteristics of the cardiovascular system for each person, it is reasonable to develop an identity recognition method based on these data. In this study, we propose an Echo State Network-based identity recognition method with CBP data. In the proposed method, we divide each CBP series data into several CBP segments. Then we use a Bidirectional Echo State Network to transform the input segments into high-dimensional reservoir states. Finally, we compute the identity recognition results in an aggregation mode. To evaluate the proposed method, we performed person identification tasks using ten sub-datasets sampled from a large-scale CBP dataset. Our proposed method achieved higher recognition accuracy than other relevant methods in spite of its relatively low computational cost on segment-by-segment and aggregated recognition tasks, respectively.