Sensitivity and Variance Analysis of Arterial Pressure Transfer Dynamics Estimated from Adaptive Multi-Channel System Identification

This paper analyzes the parametric sensitivity and asymptotic variance of the arterial pressure transfer dynamics estimated from the adaptive multi-channel (AMC) system identification technique (Hahn, 2006), a method devised to adaptively reconstruct the aortic blood pressure waveform in response to changes in the cardiovascular physiologic conditions. The parametric sensitivity analysis indicates that the pressure transfer dynamics is dominated by the wave transmission time delays at high frequency, which is confirmed by the asymptotic variance analysis, showing that the corresponding variance of the pressure transfer dynamics can be attenuated if accurate wave transmission time delays are provided. The relative parametric identifiability is then assessed by analyzing the eigenstructure of the predictor sensitivity matrix. Finally, the effect of data length on the parametric variance is also provided as a guideline for the trade-off between computational efficiency and the level of confidence.