Non-invasive estimation of brachial pressure waveforms via oscillometric transfer function analysis using cardiovascular hardware simulator

Abstract Cuff oscillations are dynamically distorted relative to arterial pressure. We investigated reconstruction of normalized brachial arterial pressure waveforms using a cardiovascular hardware simulator with six upper-arm phantom configurations. Transfer functions were identified from random excitation and evaluated on separate pulsatile recordings at 60–96 beats/min and nominal cuff pressures of 40, 100 and 160 mmHg. Eighteen individual transfer functions (ITFs) were estimated. Generalized transfer functions (GTFs) were constructed by equally weighted complex averaging; leave-one-phantom-configuration-out evaluation excluded the test configuration. On the same 72 conditions, ITF reconstruction yielded Pearson correlation r = 0.9865 ± 0.0102 and normalized root-mean-square error (nRMSE) = 0.0528 ± 0.0213, whereas held-out GTF yielded r = 0.9725 ± 0.0192 and nRMSE = 0.0713 ± 0.0235 (mean ± standard deviation). A phenomenological standard-linear-solid model fitted the complex responses with R 2 = 0.9628 ± 0.0161. Exploratory analysis of peak-height ratios identified residual reconstruction errors and sensitivity to peak-detection settings. These results support normalized waveform reconstruction in the tested simulator; they do not establish system linearity, absolute blood-pressure accuracy or clinical generalizability.