Session: 02-05-01: Data Driven Design

Paper Number: 142930

142930 - The Design of Biodegradable Scaffolds Using a Structural Descriptor and Surrogate Modeling 

Regenerative medicine is a field in medical sciences that focuses on the regrowth of tissue in the human body. Biodegradable scaffolds are important to regenerative medicine in that they provide an amicable environment for tissue regrowth. To better understand how scaffold geometry affects the degraded material properties of the scaffold, it is important to establish a structure-property (SP) linkage between the scaffold geometry and the final degraded properties. However, establishing SP relationships for scaffold design is challenging due to the complexity of the three-dimensional porous scaffold geometry. The complexity requires high-dimensional geometric descriptors. The training of such a SP surrogate model will need a large amount of experimental or simulation data. In this work, a schema of constructing SP relationship surrogates is developed to predict the degraded mechanical properties from the initial scaffold geometry.

The core of this work is divided into three tasks. Task 1: Develop a structural descriptor and generate a surrogate model for the SP relationship. Task 2: Develop a set of simulations to accurately predict the degraded material properties of a biodegradable scaffold. Task 3: Apply task 1 and task 2 to polycaprolactone (PCL) and develop an SP relationship.

To accomplish task 1, a new structure descriptor, the extended surfacelet transform (EST), is proposed to capture important details of pores associated with the degradation of scaffolds. The EST is the triple integral of a volumetric descriptor multiplied by a property of interest. Once the geometric data has been transformed into the EST domain, the EST data is then further enhanced with principal component analysis to reduce the high-dimensional EST data into a low-dimensional representation. From here, the principal component scores are used in a multiple linear and Gaussian process model to predict the material properties of the biodegradable scaffold based on the initial geometry. To accomplish task 2, a kinetic Monte Carlo biodegradation model is developed, verified, and used to simulate the biodegradation of the scaffolds to create the training data for the surrogate models. Finite element analysis is then used on these degraded scaffolds to capture the degraded materials properties of the scaffolds. To accomplish task 3, this schema is then demonstrated with the design of PCL biodegradable scaffolds by connecting the initial scaffold geometry to the degraded compressive modulus. We use a spherical descriptor in the EST with the molecular weight of each voxel as the property of interest. Using 250 training scaffolds, we show the successful creation of an SP relationship for PCL using this schema with a linear and GP surrogate.

Presenting Author: Jesse Sestito Valparaiso University

Presenting Author Biography: Dr. Jesse Sestito is an assistant professor in the Mechanical Engineering and Bioengineering department at Valparaiso University. His primary area of research focuses on multiscale modeling which includes quantum mechanics, molecular dynamics, finite element analysis, kinetic Monte Carlo, and Lattice Boltzmann simulations. He also does research in engineering design optimization using machine learning techniques such as multiscale Bayesian optimization. He has used these techniques to investigate additive manufacturing, radiation damage on materials, and biodegradation of polymeric scaffolds.

Authors:

Jesse Sestito Valparaiso University
Tequila A. L. Harris Georgia Institute of Technology
Yan Wang Georgia Institute of Technology