Abstract
Among various 3D bioprinting methods, the extrusion-based approach stands out for its ability to achieve high cell release rates and construct intricate scaffold structures. However, the use of synthetic semi-solid polymers or natural hydrogels with shear-thinning properties requires ongoing research into rheological properties, especially hydrogel viscosity. Researchers are exploring hybrid hydrogels, a combination of various materials, to ensure scaffold shape fidelity and cell viability. Current practices involve extensive experimentation to achieve the required viscosity for smooth hydrogel release through the nozzle, a process often resource-intensive and time-consuming. Addressing this challenge, computational methods, particularly machine learning, are gaining attention for fine-tuning process parameters and optimizing bio-ink components. This study adopts a decision tree-based machine learning method, demonstrating its efficacy in predicting bioink viscosity for 300 combinations of shear rates and Alginate, Carboxymethyl Cellulose (CMC), and Tempo Mediated Nano-fibrillated Cellulose (TO-NFC) material compositions. The inference model has been trained using 75% of the initial data and tested the model for the rest of the unused 25% of data. The model exhibits excellent accuracy, highlighting its potential to significantly reduce trial-and-error experiments. This approach offers a streamlined and efficient bioprinting process, paving the way for further innovations in the dynamic field of 3D bioprinting.
