Oosterlinck, Wouter

Background: Objective performance assessment is essential in surgical skill training, yet current methods are labor-intensive and focus on observing the trainee rather than the end-product of the procedure. Machine learning (ML) methods offer reproducible feedback but have mainly relied on kinematic or video data, often reducing assessment to binary or ternary classification. Our objective was to compare ML regression models predicting expert-assigned scores of vascular anastomoses from computational fluid dynamics (CFD) features of the final product. Additionally, we aimed to assess biomechanical plausibility of predictions. Methods: A total of 146 participants performed 419 end-to-side anastomoses on case-specific three-dimensional (3D) printed simulators. Anastomoses were digitized via 3D scanning, ranked by experts, and characterized using CFD-derived hemodynamic features. These served as input for linear models (Ridge, Partial Least Squares), support vector machines, and tree-based ensembles (Random Forest, Extremely Randomized Trees, and Extreme Gradient Boosting [XGBoost]), evaluated using 10-fold nested cross-validation with genetic hyperparameter optimization. Results: Inter-rater reliability of expert indicated strong agreement (intraclass correlation coefficient ICC3k = 0.846). XGBoost achieved the lowest mean root mean squared error of 0.758 (95% bootstrap CI: 0.722–0.799) and a coefficient of determination (R2) of 0.673 (0.617–0.725), with the most stable performance across folds. Shapley additive explanations (SHAP) identified the wall shear stress gradient, transverse wall shear stress, and maximum pressure as the most influential features—variables associated with intimal hyperplasia and atherosclerotic remodeling. Conclusions: Tree-based ensemble methods, particularly XGBoost, effectively modeled biomechanical properties against expert scores. Combining CFD and ML can provide reproducible, mechanistically relevant feedback in vascular surgical skill training.

Abstract Intro Energy harvesters that convert biomechanical energy into electrical power could be used for implantable devices, providing a near infinite energy solution. While various energy harvesters with piezoelectric or triboelectric elements have been designed and applied to the cardiac environment, none have been successfully integrated into leadless pacemakers due to their size constraints or insufficient energy generation. Purpose This study aimed to develop a proprietary inertial piezoelectric energy harvester (PEH) capable of generating the energy required for a self-sustaining leadless pacemaker from the beating of the heart. Method Human heart dynamics were tested using ultra-fast MRI scans of 20 patients, which were analyzed for displacement of the right ventricular (RV) apex, mid- to low septum, and the right atrial (RA) walls. The data was used to design the PEH, which was evaluated in thirteen (13) male swine. In addition, epicardial MEMS-based sensors recorded the acceleration, angular positions, and gravity in 3 directions. The sensors were placed at the apex of the RV, LV Apex, LV free wall, and the RA using adhesive means. The data collected from the animals was used to calculate the power produced by the PEH in the different regions of interests and to create a bench model to optimize the PEH design further. The PEH was then tested in-vivo in an ovine model. Results The heart contractility creates harmonic acceleration in the heart tissue, transmitting this vibration into the PEH, resulting in the harvester acceleration of 0.75 g (±0.25g). When the PEH is subject to vibrations, the PEH oscillates using kinetic forces exerted on the piezoelectric beam through a proof mass, transforming these mechanical forces into electric energy. The electrical energy produced using this PEH is 6µW (± 2µW), which is sufficient to power a leadless pacemaker, providing pacing therapy of 100% pacing, at 60 BPM, with heart impedance of 600 (±200), and a pacing voltage reaching 2V, with sufficient excess energy to maintain charge of a secondary battery used for communication and advanced pacemaker functions. This was then tested in a pre-clinical setup, using an ovine model (results to be presented in the future). Conclusion This work demonstrates that a proprietary inertial PEH can use the contraction energy of the heart to generate electrical energy. The design of the PEH was based on the analysis of human heart kinetics using MRI scans and in-vivo recordings from pigs’ hearts. This design of an effective energy harvester can produce enough energy to power a leadless pacemaker. The design can fit in a small capsule that can be used in a leadless pacemaker that will be implanted in the RV providing sensing and pacing as well as daily communication enabling advanced pacemaker functions. The PEH is currently being tested in an ovine model, demonstrating its safety and performance.