Automated triage parameters estimation from ECG

Abstract

Low-cost biosensors combined with low-cost portable devices can be very useful in time critical situations of mass casualties, when fast triage procedure must be attained. A methodology that uses ECG to derive the vital parameters (heart rate and respiratory rate) needed for the triage procedure is presented and it is aimed to leverage affordable low-cost equipment that can be easily utilized by urgent medical units or even volunteers in events of considerable number of injured civilians. The methodology relies on selected well-known and published algorithms for heart rate and respiratory rate derivation from a given ECG signal. It consists of methods for R-wave detection, kurtosis computation, smoothing, and finding peaks. The proposed approach is shown to offer a good trade-off between the accurate measurement of the parameters and their fast derivation. It has been evaluated by using a publicly available database. Its robustness is measured in terms of accuracy estimation, showing a sensitivity of 0.87 for heart rate and 0.74 for respiratory rate, a sensitivity of 0.76 considering the triage process and an average-case execution time of 0.02 seconds, making it suitable for real-time applications.