Abstract 272: Developing a Hypovolemia Monitoring System by Integrating Physiological Measures and P-QRS-T Waves
Introduction: Early detection of hemorrhage has a critical impact on patient survival, making it vital to quickly and accurately identify those patients at risk for developing overt shock. Current standard physiological measures do not allow for the early detection of blood loss. This study investigated the integration of physiological measures and P-QRS-T wave changes from the ECG for determining the severity of hypovolemia.
Methods: The approach analyzing both physiological measures and P-QRS-T wave changes was tested with data collected on 86 human subjects undergoing lower body negative pressure (LBNP) as a hemorrhage mimetic. LBNP was applied at 5-min intervals of −15, −30, −45, and −60 mm Hg, followed by decrements of −10 mm Hg until the onset of cardiovascular collapse. LBNP levels were retrospectively divided into 3 classes of hypovolemia (mild, moderate, and severe) based on the level that cardiovascular collapse occurred. Five physiological measures (heart rate, pulse pressure, shock index, pulse pressure variation, and systolic pulse variation) and the P-QRS-T intervals (PQ, QS, QT, PR, ST) were measured from beat-to-beat arterial blood pressure waveforms and ECG. New features were extracted using non-overlapping 60-sec sliding windows. Accuracy, sensitivity and specificity of the integrated approach in detecting the severity of hypovolemia was tested using LibSVM machine learning with ten-fold cross-validation and compared to those obtained using physiological measures and P-QRS-T intervals alone.
Results: All three approaches were tested on 1810 sample cases using LibSVM.
Results: showed 73.4% accuracy in detecting the severity of LBNP-produced hypovolemia using physiological measures alone, 82.5% using P-QRS-T intervals alone, and 88.9% when integrating all measures. The sensitivity and specificity of the integrated approach were 88.6% and 88.4%, respectively.
Conclusions: Integration of physiological measures and P-QRS-T intervals were used in detecting hypovolemia severity produced by LBNP. Combining the two sources improved accuracy compared to the use of either source alone. The integrated approach may prove useful in rapidly determining the degree of volume loss from hemorrhage, allowing for earlier decision making.
- © 2010 by American Heart Association, Inc.