Related Experiment Video
Updated: Oct 30, 2025
08:27
Design and Validation of a Volumetric-extrusion Bioprinter for Bioprinting of Soluble Basement Membrane Extract for Translational Research
Published on: March 28, 2025
460
Machine Assisted Experimentation of Extrusion-Based Bioprinting Systems
Shuyu Tian1, Rory Stevens1, Bridget T McInnes2
1Department of Chemical and Life Science Engineering, Virginia Commonwealth University, Richmond, VA 23284, USA.
Micromachines
|July 2, 2021
Summary
Machine learning models trained on bioprinting literature data can predict extrusion-based bioprinting (EBB) outcomes. These models offer a faster, more accessible approach to optimizing EBB parameters like cell viability and filament diameter.
Area of Science:
- Biomaterials Science
- Biofabrication
- Computational Biology
Background:
- Extrusion-based bioprinting (EBB) parameter optimization is typically experimental, requiring significant time and resources.
- This experimental approach is often difficult to replicate across different laboratories.
- Existing methods lack a systematic, data-driven approach for predicting bioprinting outcomes.
Purpose of the Study:
- To develop and evaluate machine learning (ML) models for predicting EBB parameters using published literature data.
- To assess the efficacy of regression and classification ML models in forecasting cell viability and filament diameter.
- To investigate the potential of ML to predict optimal extrusion pressure for desired cell viability.
Main Methods:
- Collected and curated data from published literature on EBB experiments.
- Developed and trained regression-based and classification-based ML models.
- Compared model performance using general literature data versus single-source literature data.
- Evaluated model accuracy in predicting cell viability, filament diameter, and extrusion pressure.
Main Results:
- ML models trained on extensive literature data successfully captured trends in cell viability, filament diameter, and extrusion pressure.
- Regression models trained on larger datasets provided more accurate cell viability predictions for untested material concentrations.
- Classification models achieved up to 70% accuracy for cell viability but showed limited sensitivity to input parameter changes.
- Models trained on diverse literature data outperformed those trained on single-source data.
Conclusions:
- Machine learning offers a powerful, data-driven alternative to traditional experimental optimization of EBB parameters.
- ML models can accelerate bioprinting experimental design and improve prediction accuracy for critical printing outcomes.
- Further development of ML approaches holds significant potential for advancing the field of bioprinting.
