Feedback Control of Bioelectronic Devices Using Machine Learning

Bioelectronic devices have shown their unprecedented potential in a wide range of biomedical applications due to their vast functionality. To fully unleash their potential, bioelectronic devices should be able to precisely respond to real-time changes in the environment to drive biological systems' response towards the desired goals. However, controlling the biological systems' response with bioelectronic devices is challenging due to the presence of uncertainties, stochasticity, unmodeled dynamics, and complex nonlinearities. Here, we demonstrate the promise of leveraging tools in synthetic biology with an online machine learning (ML)-based feedback controller to achieve a precise spatiotemporal response of biological systems using bioelectronics driven by an adaptive external “sense and respond” learning algorithm. The proposed ML-based feedback controller leverages the learning capability and low computational complexity of a class of artificial neural networks called the Radial Basis Function (RBF) network. Lyapunov analysis is provided to prove stability of the proposed controller. The satisfactory performance of the proposed method is experimentally validated by maintaining media pH using bioelectronic proton pumps.