Accepted_test

Technologies of 3D bioprinting currently in rapid development provide new opportunities for regenerative medicine and tissue engineering. Moreover, lab-made three-dimensional biomodels have become an important tool for studying disease dynamics and testing of new remedies in environments that closely resemble natural ones. However, the need for search of optimal printing parameters remains main challenge in creation of biocompatible scaffolds – composite porous structures made of hydrogel and infused with cells. Application of active machine learning methods allows to reduce the number of experimental iterations needed for achieving required quality and effectiveness of bioprinting. We develop a model based on active machine learning for optimization of parameters for printing biocompatible scaffolds with extrusion 3D bioprinter. The model selects the most informative experiments for improvement of printing quality based on its internal estimation of prediction confidence. After that we change the printing parameters, such as syringe temperature, printing speed and hydrogel composition during the experiment based on the recommendations given by the model. Application of the active learning-based model allows to improve the quality of biocompatible scaffolds while reducing the number of required experimental iterations. This approach speeds up parameter selection for accurate printing. The results of the current research may be applied towards development of individualized biomedical solutions and their translation into clinical practice.