Optimizing the Sewing Position of Flexible Piezoelectric Energy Harvesters on a Three-Dimensional Deforming Heart: An Integrated Approach Using Finite Element Analysis, Deep Learning, and Theoretical Modeling

Abstract

Flexible piezoelectric energy harvesters (FPEHs) have attracted tremendous attention due to their potential applications in the field of biomedicine, such as powering implantable devices. Despite observations in numerous in vivo experiments that the electrical output of FPEHs varies considerably with sewing positions during energy harvesting from heartbeats, optimal sewing positions have not been thoroughly investigated. In this article, an approach that integrates finite element analysis (FEA), long short-term memory (LSTM) deep learning method, and theoretical modeling was proposed to investigate the impact of the sewing position on the harvest performance of the FPEH, utilizing real three-dimensional heart deformation data as the end-to-end displacement load for the FPEH. The results reveal that the sewing positions have a significant influence on the electric output performance of the FPEH. The optimal sewing position was identified near the posterior interventricular groove on the upper part of the left ventricle, with a corresponding optimal resistance value of 8 MΩ and an output power of 122.9 nW. Additionally, five suggested sewing positions across different regions of the heart's surface were provided for clinical application. The methodology that integrates FEA, deep learning approach, and theoretical modeling in this article can be extended to determine the optimal position for the flexible devices patching on other irregular and deforming surfaces.