Noninvasive Predictive Model for Rapid Hemodynamic Optimization During Mechanical Circulatory Support

Mechanical circulatory support (MCS) devices are standard therapy for advanced heart failure, but modern systems lack standardized methods for setting critical parameters such as rotational speed and flow rate. We developed an integrated model to support MCS parameter adjustment and provide continuous hemolysis and hemodynamic monitoring. The model combines lumped parameter modeling (LPM) and reduced order modeling (ROM) derived from high-resolution computational fluid dynamics of the device. Sensitivity analysis and Bayesian algorithms were used to ensure efficient model calibration and to identify and quantify the most influential parameters driving model outputs. Retrospective data from clinical use of the magnetically levitated centrifugal extracorporeal ventricular assist device MoyoAssist (magAssist, Suzhou, China) were analyzed. The model efficiently fit clinical data (n = 11) using five initial and two subsequent parameters. It provided key hemodynamic information such as left ventricular elastance, pressure-volume (P-V) loops, and hemolysis risk. The model revealed a relationship between rotational speed and cardiac index (CI), identifying a safe adjustment range with a total cardiac index (CI_total) >2.2 and native cardiac index (CI_heart) >0.25. This demonstrates feasibility for rapid P-V loop and ventricular elastance calculation after deployment, providing a basis for defining target CI values and recommending speed adjustments to optimize MCS use. https://links.lww.com/ASAIO/B889.