Robust computation of pulse pressure variations

Evidence of arterial pulse pressure variations caused by cardio-pulmonary interactions, and their connection to volume status via the Frank-Starling relationship, are well documented in the literature. Computation of pulse pressure variations from arterial pressure measurements is complicated by the fact that systolic and diastolic peaks are not evenly spaced in time. A robust, structurally uncomplicated, and computationally cheap algorithm, specifically addressing this fact, is presented. The algorithm is based on the Lomb-Scargle spectral density estimator, and ordinary least squares fitting. It is introduced using illustrative examples, and successfully demonstrated on a challenging porcine data set. Keywords— pulse pressure variation, arterial blood pressure, nonuniform sampling, frequency estimation Highlights • A method for robust, fast computation of arterial pulse pressure variations, is presented. • The method is based on the Lomb-Scargle periodogram and least squares regression. • The algorithm is particularly suitable for closed-loop control, and other time-critical applications. • A porcine dataset with sudden hemodynamic changes is used to demonstrate feasibility.