Predicting printability in suspended bioprinting using a rheology-informed hierarchical machine learning approach

Suspended bioprinting has emerged as a promising method for overcoming the limitations of conventional extrusion-based bioprinting, enabling the creation of complex tissue constructs with improved resolution and shape fidelity. This technique utilizes a support bath to preserve the structural integrity of bioinks during deposition, allowing for the precise printing of low-viscosity materials. However, optimizing printability remains a significant challenge due to the absence of standardized methods and the complex interactions between bioink properties, support bath characteristics, and printing parameters. This study introduces a novel approach integrating suspended bioprinting with a rheology-informed hierarchical machine learning (RIHML) model to predict key printability factors such as axial resolution, horizontal resolution, and z-axis positional errors. A comprehensive dataset was generated by varying rheological properties and printing conditions to train and validate the RIHML model. The results show that the RIHML model outperforms conventional machine learning models, including support vector regression and concentration-dependent model, in predictive accuracy. This approach addresses critical challenges in suspended bioprinting, offering a scalable solution for improving printability, enhancing cost-effectiveness, reducing time consumption, and boosting the precision and reproducibility of tissue-engineered scaffolds.