A Preliminary Study on the Usage of a Data-Driven Probabilistic Approach to Predict Valve Performance Under Different...

Area of Science:

• Biomedical Engineering
• Cardiovascular Research
• Medical Device Technology

Background:

• Predicting complications after aortic valve replacement (AVR) is vital for pre-procedural planning.
• High transvalvular pressure gradients (DP) can indicate potential complications.
• Data-driven models can estimate DP without extensive experiments.

Purpose of the Study:

• To generate logistic regression models for predicting the probability of transvalvular pressure gradient (DP) exceeding 20 mmHg after AVR.
• To assess the hemodynamic performance of SAPIEN 3 and Magna Ease aortic valves under various physiological conditions.
• To validate the predictive accuracy of the developed models against experimental data.

Main Methods:

• Hemodynamic assessment of SAPIEN 3 and Magna Ease valves under pulsatile flow.
• Simulated a range of systolic blood pressures (100-180 mmHg), diastolic blood pressures (40-100 mmHg), and heart rates (60, 90, 120 bpm).
• Developed logistic regression models to predict DP > 20 mmHg and validated using receiver operating characteristic (ROC) curves and area under the curve (AUC).

Main Results:

• Probabilistic predictive models for DP > 20 mmHg were generated for both SAPIEN 3 and Magna Ease valves.
• The SAPIEN 3 model achieved an AUC of 0.9465, and the Magna Ease model achieved an AUC of 0.9054, indicating high accuracy.
• Experimental and model-predicted DP probabilities showed good agreement.

Conclusions:

• The developed data-driven models accurately predict the probability of high transvalvular pressure gradients for specific aortic valves.
• These models represent a significant step towards improving pre-procedural planning for aortic valve replacement.
• The findings support the use of predictive modeling for assessing valve reliability in AVR.