Noninvasive pulsatile flow estimation for an implantable rotary blood pump

Area of Science:

• Biomedical Engineering
• Cardiovascular Devices
• Fluid Dynamics

Background:

• Implantable rotary blood pumps (iRBPs) are crucial for treating heart failure.
• Accurate pulsatile flow estimation in iRBPs is challenging but vital for patient monitoring and device optimization.
• Existing methods often require invasive measurements, limiting their clinical applicability.

Purpose of the Study:

• To develop and validate a noninvasive method for estimating pulsatile flow in iRBPs.
• To assess the efficacy of system identification techniques using pump feedback signals and hematocrit (HCT).
• To provide a practical tool for real-time flow monitoring in pulsatile environments.

Main Methods:

• Acquired pulsatile flow data using six fluid solutions (20-50% HCT) in an in vitro mock circulatory loop.
• Developed autoregressive with exogenous input (ARX) models using pump speed, power, and HCT as inputs.
• Evaluated three ARX model structures, including those incorporating HCT and non-pulsatile flow estimates.

Main Results:

• Optimized ARX models demonstrated high accuracy in flow estimation across the pump's operating range (-2 to 12 L/min).
• Structure II models achieved a minimum mean flow error of 5.49% (0.258 L/min) on unseen data.
• Structure III models yielded a minimum mean flow error of 5.77% (0.270 L/min), validating their performance.

Conclusions:

• The developed system identification models offer a practical and accurate noninvasive approach to pulsatile flow estimation in iRBPs.
• These models can enhance patient monitoring and improve the management of patients with iRBPs.
• The findings support the integration of these noninvasive estimation techniques into clinical practice.