Physiologic Data-Driven Iterative Learning Control for Left Ventricular Assist Devices.

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Control Systems

Background:

• Continuous flow ventricular assist devices (cfVADs) are crucial for end-stage heart failure but often operate at fixed speeds, limiting physiological response and potentially causing adverse events.
• Current cfVAD controllers lack pulsatility, a feature thought to be beneficial for cardiovascular system function.

Purpose of the Study:

• To develop and evaluate a physiological data-driven iterative learning controller (PDD-ILC) for cfVADs.
• To achieve physiological, pulsatile, and treatment-driven cfVAD response.
• To investigate pulsatility maximization and left ventricular stroke work (LVSW) minimization using copulsation and counterpulsation modes.

Main Methods:

• Developed a PDD-ILC and a physiologic pump flow proportional-integral-derivative controller (PF-PIDC).
• Tested controllers in an in-silico environment under various physiological conditions.
• Compared PDD-ILC and PF-PIDC against constant speed (CS) control and existing model-based iterative learning control approaches.

Main Results:

• Both PDD-ILC and PF-PIDC demonstrated highly accurate tracking of reference pump flow trajectories, outperforming existing methods.
• Controllers successfully achieved predefined treatment objectives (pulsatility maximization and LVSW minimization).
• PDD-ILC and PF-PIDC resulted in improved hemodynamics and preload sensitivities compared to CS support.

Conclusions:

• The PDD-ILC offers a promising approach for achieving physiological and pulsatile cfVAD support.
• Developed controllers improve cfVAD therapy by enabling tailored, dynamic responses and better patient outcomes.
• This data-driven, iterative learning approach advances cfVAD control strategies for heart failure management.