Cardiac output estimation from arterial blood pressure waveforms using the MIMIC II database

Abstract

The effect of signal quality on the accuracy of cardiac output (CO) estimation from arterial blood pressure (ABP) was evaluated using data from the Multi-Parameter Intelligent Patient Monitoring for Intensive Care (MIMIC) II database. Thermodilution CO (TCO) was the gold standard, and a total of 121 records with 1,497 TCO measurements were used. Six lumped-parameter and systolic area CO estimators were tested, using ABP features and a robust heart rate (HR) estimate. Signal quality indices for ABP and HR were calculated using previously described metrics. For retrospective analysis, results showed that the Liljestrand estimator yielded the lowest error for all levels of signal quality and for any single estimator when using five or more calibration points. Increasing signal quality decreased error and only marginally reduced the amount of available data, as a signal quality level of 90% preserved sufficient data for almost continuous CO estimation. At the recommended signal quality thresholds, the lowest gross root mean square normalized error (RMSNE) was found to be 15.4% (or 0.74 L/min) and average RMSNE was 13.7% (0.71 L/min). Based on these results, a linear combination (LC) of the six CO estimation methods was developed and proved superior to all other methods when up to 13 TCO calibration values were used. The clinical utility of the CO estimates were examined by correlating changes in four vasoactive medication doses with corresponding changes in estimated resistance, which was derived from mean ABP and estimated CO.

(cont.) Both the Liljestrand estimator and the LC estimator were used to estimate CO. Regression analysis failed to show a clear correlation between dose level and estimated resistance for either estimator except for neosynephrine, revealing the limitations of current SQI methods in ensuring signal fidelity. Examples of types of non-physiological or artifactual ABP waveforms are shown, and a potential damping detection method is proposed.