A NUMERICAL AND EXPERIMENTAL FRAMEWORK FOR NON-INVASIVE AND CONTINUOUS MONITORING OF TRANSCATHETER HEART VALVES

Transcatheter aortic valve implantation (TAVI) has become the standard treatment for aortic stenosis, with increasing interest in its application to younger patients. Nonetheless, the long-term durability of transcatheter heart valves (THVs) remains a significant concern, as the progressive deterioration of the biological leaflets can impair their functionality. Conventional diagnostic imaging is insufficient for continuous functional monitoring, underscoring the necessity for non-invasive solutions for the early detection of valve dysfunction.This thesis develops an integrated computational and experimental framework for designing and characterizing a non-invasive THV monitoring system utilizing photoplethysmographic (PPG) sensors and machine learning algorithms. A bidirectional fluid-structure interaction (FSI) analysis was conducted on an idealized aortic vessel model, both before and after the virtual deployment of a SAPIEN 3 Ultra valve, to evaluate the integration of optical fibers and PPG sensors. The simulations facilitated the identification of optimal sensor placement and demonstrated that pulse wave velocity and pulse transit time are effective metrics for assessing the transvalvular pressure gradient. Experimental assessment of the framework was performed using a 3D-printed aortic phantom and a pulsatile flow circuit, showing good agreement between numerical predictions and experimental data. Finally, a proof-of-concept system was developed in which a self-expanding Evolut FX valve was subjected to multiple hemodynamic scenarios generated through Latin Hypercube Sampling. PPG signals and flow variables were combined with machine learning models to predict the geometric orifice area of the valve and classify leaflet mobility, demonstrating high predictive capability and accurate discrimination between normal and pathological conditions. Overall, this thesis establishes a unified methodological framework that integrates computational biomechanics, experimental investigation, and machine learning techniques for intelligent monitoring of valvular prostheses. This approach offers a promising tool for the early detection of valve dysfunction and the improvement of long-term management in patients undergoing TAVI, laying the groundwork for the development of smart prosthesis monitoring systems and the safe extension of the procedure to younger populations.