Analyse de signaux de ballistocardiogramme pour la détection des événements cardiorespiratoires

Abstract

The new generation of connected mattresses opens up interesting avenues for monitoring vital signs in clinical settings. This paper proposes a study of the utility and validity of such technology in a population with cardiorespiratory problems in a clinical setting. Discreet and continuous monitoring of the respiratory rate (RR) is essential to provide better diagnostic when using smart objects. As a result, new intelligent mattresses are increasingly focusing on respiratory rate (RR). In our work, we validate the performance of a microbend FOS mattress on 80 subjects admitted to the cardiac intensive care unit (CICU) by estimating the RR from their ballistocardiograms (BCG). Our study proposes a new estimator of the RR, based on harmonic plus noise (HNM) models and compares it to two known estimators (namely MODWT and CLIE). The objective is to examine, using a representative data set, the accuracy of these methods and of the smart mattress itself. The results of applying these three estimators to the BCG show that the MODWT is more accurate with an average absolute mean absolute error (MAE) of 5.72 ± 6.47 BPM. The CLIE estimator was less efficient with an MAE of 10.90 ± 8.35 BPM. The HNM estimator shows encouraging results of 8.45 ± 8.22 BPM. Bland-Altman analyses revealed that each of these estimation methods works better for slow RRs, but that the error increases linearly with the frequency of the RR. This means that the smart mattress works well within a standard RR range of 10 20 breaths per minute (BPM) but becomes less accurate with a wider range of estimation. These results highlight the need to improve the performance of these sensors for wider ranges of vital signs. In short, this project proposes a much broader database than the previous ones on a population with different cardiorespiratory disorders. Moreover, it validates that a smart mattress can function, under good conditions, on this same population.

Date	24 Oct 2020
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Jacques A. de Guise (Supervisor) & Neila Mezghani (Co-supervisor)