Pulse Wave Velocity Estimation in a Controlled In Vitro Vascular Model: Benchmarking Machine Learning Approaches.

Area of Science:

• Biomedical Engineering
• Fluid Dynamics
• Materials Science

Background:

• Accurate estimation of arterial stiffness and pulse wave velocity (PWV) is crucial for cardiovascular health assessment.
• In vitro models provide a controlled environment to develop and validate novel measurement techniques.

Purpose of the Study:

• To evaluate the feasibility of estimating stiffness parameters and PWV in artificial silicone vessels.
• To benchmark a neuro-fuzzy inference system against other machine learning models for hardness prediction.
• To assess PWV estimation accuracy using hardness-derived elasticity and the Moens-Korteweg equation.

Main Methods:

• Utilized a controlled in vitro circulatory setup with silicone vessels of varying hardness and thickness.
• Extracted synchronized pressure and capacitive waveforms, identifying fiducial points and engineered features.
• Employed a Sugeno-type adaptive neuro-fuzzy inference system (ANFIS) for hardness prediction, compared with regression and ML/DL models.
• Calculated PWV using hardness-to-elasticity conversion and Moens-Korteweg formulation, comparing against a reference PWV.

Main Results:

• The proposed pipeline demonstrated strong agreement with reference labels and measurements under controlled in vitro conditions.
• The ANFIS model showed competitive performance in predicting vessel hardness levels.
• PWV estimates derived from the developed methods aligned well with reference measurements.

Conclusions:

• The study successfully validated a pipeline for estimating stiffness parameters and PWV in an in vitro setting.
• The findings support the potential of this approach for cardiovascular research, pending further in vivo validation.
• Future work requires external validation, calibration, and robustness testing for physiological applications.