Abstract
Machine learning (ML) dates as far back as 1959, and focuses on the use of large data sets to identify underlying patterns and/or predict future behavior. Tissue engineers and other scientists are well acquainted with traditional statistical analyses, which evaluate formalized relationships between independent and dependent variables, typically under a specific set of assumptions. ML models, in contrast, attempt to identify less explicitly defined relationships in the data by iteratively assessing—that is, learning—how variables interact with each other. Thus, ML models are well equipped to identify complex associations between multiple features of a data set. In practice, ML models are particularly suitable for data that involve heterogeneous inputs or a large number of variables, which often applies to the chemical and biological contexts of tissue engineering research.
ML technology has become increasingly accessible to scientists, engineers, and professionals of all backgrounds. Consumer-grade computers are now largely capable of running ML analysis, and ML tools also come packaged with popular analytical software such as MATLAB and SAS. As a result, ML has accelerated processes of scientific discovery and data analysis across multiple disciplines, ranging from health care to drug development and structural biology.
Tissue engineering and other biomedical fields have increasingly adopted ML strategies to investigate complex phenomena such as structure–function relationships in biomaterials and optimization of bioengineered processes. In this special issue, we highlight a selection of research at the forefront of this evolving intersection between ML and tissue engineering. The first article in this issue provides a broad overview of the emerging applications of ML to tissue engineering research, including biomaterial development, construct fabrication, and tissue biology.
The second article discusses original research on the usage of convolutional neural networks to automatically classify cardiomyocyte content during induced pluripotent stem cell differentiation. The third article describes the development of a novel computational model to simulate three-dimensional vascular matrix architecture in the human thyroid using clinically acquired ultrasonographic images. Finally, the fourth article provides an overview of the applications of ML to bioelectronics, including considerations in material design, device fabrication, and the analysis of biosensor data.
This timely issue highlights the scientific and biomedical potential emerging at the intersection of ML and tissue engineering. The guest editors hope that this issue will spark interest in the myriad of potential applications for ML analysis, particularly as scientific data become increasingly complex. Ultimately, ML may provide an invaluable tool kit for tissue engineers to accelerate data exploration, pattern recognition, and clinical translation.