Abstract 200: Computer-Assisted Monitoring of Hemodynamic Condition During a Simulated Experiment of Hypovolemia Using Lower Body Negative Pressure
Background: Real-time physiologic status monitoring of hemorrhaging war-fighter or civilian from the point of injury and along ensuing echelons of care in austere environments pose significant challenges. These challenges range for those arising from the environment to the physical profiles of monitors and the types of measures they make.
Objective: To develop an automated clinical support system capable of using low level physiologic signals such as the electrocardiogram (ECG) to monitor hemodynamic compensation in response to injury or illness.
Methods: An adapted low profile portable and wearable armband monitor was used to collect prolonged measurement of a single lead ECG from volunteer human subjects undergoing lower body negative pressure (LBNP) as a model of human hemorrhage. Fifty-five subjects underwent 5-minute intervals of progressive LBNP until hemodynamic decompensation at which time LBNP was halted. Using several advanced signal processing techniques including novel applications of the Stockwell Transform and L-1 Norm optimized taut-string theory, hidden underlying bio-markers were extracted from the ECG signal which was then used to develop a predictive model using support vector regression using radial basis kernel to predict the severity of blood loss with respect to LBNP levels.
Results: Using by-subject cross validation for each of the 55 subjects, the Pearson’s correlation coefficient was calculated between the algorithm’s estimates and the actual LBNP levels for each subject. High correlation between the prediction and actual levels were observed with a mean correlation coefficient across all subjects of 0.862 and σ of 0.072. The mean r2 value was 0.928 with σ of 0.04. The system predicted changes in stability prior to any significant changes in heart rate and blood pressure among subjects during LBNP.
Conclusions: The developed ECG analytic technique has provided encouraging results and has shown potential to be applicable in a variety of critical care environments and applications where detection of hemodynamic changes prior to overt decompensation would allow early intervention. This when used with wearable sensors opens up new possibilities for monitoring in different locations.
Author Disclosures: A. Belle: None. M. Spadafore: None. H. Derksen: None. K. Ward: None. K. Najarian: None.
- © 2014 by American Heart Association, Inc.