Page 58 - IJB-9-4

International Journal of Bioprinting Machine learning and 3D bioprinting

Figure 2. ML applications in bioprinting.

Figure 3. ML methods.

Hydrogel scaffolds fabricated using EBB and fibrous 1.2. Common ML methods used in bioprinting
scaffolds fabricated using EHD can mimic extracellular ML has experienced rapid progress over the past two
matrix (ECM) components from the native environment decades and has demonstrated outstanding capability
and influence cell behaviors and outcomes [11,14,16] . Numerous in pattern identification and parameter optimization
experiments have been conducted to investigate material for metal machining and printing [17,18] . As an emerging
properties, process parameters, and their effects on scaffold technology, it has the potential to streamline the current
building. However, researchers in cell biology or drug bioprinting workflow through process–material–
screening may not be available or capable of experimenting performance modeling.
with these factors. This has impeded the widespread As shown in Figure 2, the current ML applications in
adoption of these technologies across multiple disciplines.
bioprinting include material property optimization for
In addition, the concept of customized constructs to reliable printability and shape fidelity, process optimization
specifically tailor cell responses in 3D cultures has drawn with the desired fiber or droplet diameter, in situ process
ongoing research interest. Such constructs require the precise monitoring for stable fabrication and process adjustment,
control of biomaterial compositions, structural designs, and bioprinted construct optimization for better cell–
and printing technologies, which cannot be realized using microenvironment interactions. Both traditional ML
experimental or mathematical models. ML has proven its and DL methods are applied to develop prediction,
capability to model complex processes with multiperformance segmentation, and detection models .
[19]
characteristics. It is therefore introduced to systematically The traditional ML methods shown in Figure 3a use
model materials and parameters, as well as to quantitatively link extracted and selected numerical features as inputs for
process–material–performance in bioprinted constructs .
[17]
prediction tasks. Using the collected dataset or images,

Volume 9 Issue 4 (2023) 50 https://doi.org/10.18063/ijb.717