A reinforcement learning approach to the computational generation of biofabrication protocols

Biofabrication is an emerging field that identifies the set of processes required for the generation of biologically functional products with structural organization and subsequent tissue maturation processes. Biofabrication technologies have the potential to revolutionize the Regenerative Medicine RM domain, which aims to regenerate damaged tissues. Generally RM techniques are costly and time-consuming, so computational approaches can tap into this potential, facilitating innovation in biofabrication and supporting process and product quality. This Thesis project aims at the development of a software framework capable of generating optimal protocols to improve biofabrication processes. The framework will combine simulation techniques and Reinforcement Learning RL approaches to computationally generate optimal protocols for the simulated fabrication of epithelial sheets , while providing a customizable interface that can be set up with any simulator. The optimization engine uses a Deep Learning DL-based reinforcement learning algorithm, the Advantage Actor Critic A2C, which relies on a customizable neural network that adapts to the specific environment given to the engine through the interface. The potential of the framework is demonstrated through the optimization of a cell proliferation process, with both the easiness of implementation of the specific process and the potentiality of the RL approach to improve Biofabrication processes in the future.