A team of researchers has developed a novel approach for real-time estimation of cardiac output and systemic vascular resistance using only arterial blood pressure measurements. The method could provide clinicians with crucial hemodynamic information while avoiding the risks associated with more invasive monitoring techniques.

In the study, published in IEEE Transactions on Biomedical Engineering, the researchers utilized a state-space model based on the two-element Windkessel model of the cardiovascular system. By analyzing the diastolic decay of arterial blood pressure (ABP) waveforms, the researchers were able to estimate a time constant τ, which is proportional to systemic vascular resistance (SVR). From this, they derived estimates of proportional cardiac output (CO).

Among the key findings were:

• Estimates of τ and proportional CO derived from ABP alone closely aligned with those obtained using both ABP and aortic flow measurements in a swine model.
• Estimates were consistent across central, femoral, and brachial ABP measurement sites in most cases.
• The method produced predictable changes in estimated SVR and CO in response to common cardiovascular drugs.

The researchers applied their technique to data from 6 Yorkshire swine under anesthesia. ABP was recorded from femoral, brachial, and central locations, while aortic flow was measured using an ultrasonic flow probe. Various cardiovascular drugs were administered to manipulate hemodynamics.

For τ estimation, the researchers constructed a state-space model with τ as the hidden state and diastolic ABP samples as observations. A Kalman filter was used to obtain beat-by-beat estimates. Proportional CO was then derived using these τ estimates and ABP measurements.

When comparing ABP-derived estimates to those using both ABP and aortic flow, the researchers found strong correlations. For τ estimates, Pearson correlation coefficients ranged from 0.89 to 0.95 across measurement sites. For proportional CO estimates, correlations ranged from 0.79 to 0.94. The average absolute percent error (AAPE) for τ estimates ranged from 8% to 17%, while for proportional CO estimates it ranged from 7% to 13%.

The method also demonstrated consistency across ABP measurement locations. Comparing central to femoral estimates, correlation coefficients were 0.98 for τ and 0.93 for proportional CO. Central to brachial comparisons showed correlations of 0.94 for τ and 0.85 for proportional CO.

Drug response analysis revealed expected hemodynamic changes for several common cardiovascular medications. Phenylephrine administration increased estimated SVR, while dobutamine showed concomitant decreases in SVR and increases in CO. Esmolol demonstrated vasoconstrictive effects and decreased CO. Nitroglycerin produced arterial and venous dilation effects, decreasing both SVR and CO. Fentanyl showed vasodilatory effects.

The researchers noted that prominent dicrotic notches in central ABP could lead to discrepancies in estimates compared with peripheral measurements. They also detailed specific signal processing techniques used for ABP and aortic flow, including filtering and feature detection methods.

The researchers emphasized that their approach could reduce reliance on highly invasive monitoring techniques like pulmonary artery catheters. Peripheral arterial catheters, already common in critical care settings, could provide valuable hemodynamic data with lower associated risks.

Limitations of the study included the need for validation in larger datasets and human subjects. The researchers also acknowledged that their method estimated proportional rather than absolute values of SVR and CO. They plan to focus on separate estimation of R, C, and CO in future work.

The novel technique for estimating key hemodynamic parameters using only ABP measurements showed promise for improving cardiovascular monitoring in critical care settings. Further research is needed to validate its clinical utility and reliability across diverse patient populations.

Conflict of interest disclosures were not made available at time of publishing.