P-3
Modeling neural control of heart rate under a stochastic regime: Parameter distribution sampling and patient-specific model adaptation for clinical inference in the critically ill

https://doi.org/10.1016/j.jcrc.2012.01.025Get rights and content

Section snippets

Objectives

The implementation of physical modeling tools in a critical patient-care setting can provide a major access pathway to information, which may be readily recast into a clinically useful form. In particular, when a priori knowledge about system dynamics is injected into the modeling process (Guerrisi et al., IFMBE Proc., 2005:11(1). Guerrisi et al., Physiol Meas. 2009 Jan;30(1):81-100. Toschi and Guerrisi. Med Biol Eng Comput. 2008 Jul;46(7):637-48), model output as well as parameter values can

Methods

(a) Experimental data collection: invasive arterial blood pressure (ABP) and electrocardiogram waveforms were collected from a group of critically ill patients undergoing major surgery at the University Hospital Rome “Tor Vergata” using an in-house developed software tool (“Global Collect”, Labview 2009 environment) that allows simultaneous recording and real-time visualization of data from multiple patient monitors (GE S/5 Avance Carestation [GE Healthcare, Waukesha WI], Pulsion PiCCO [PULSION

Results

After introducing stochastic noise into an existing cardiovascular control model (Seidel and Herzel. Physica D. 1998, 115:145-162), the former is combined with a lumped-parameter model of the cardiovascular system (Heldt et al. Electrophysiol Ther J. 2010, 3:45-54) to generate realistic, coupled R-R interval and pressure waveforms. The probability density of R-peak occurrence is extrapolated as a functional form of model parameters and noise levels and successively used to quantify the density

Conclusion

A highly refined inferential framework is set up, which allows matching and validation of a neural blood pressure control mechanism model against in vivo data collected from critically ill patients. Clustering of most probable parameters is observed within the same subject and between a different subject, suggesting the possibility of real-time determination of a univocal connection between interpretable model parameters and underlying patient state.

References (0)

Cited by (0)

View full text