Artificial Cardiac Pacemaker Control design using Deep Reinforcement learning: A Continuous Control Approach

Abstract

Cardiac pacemaker is a device that demands very high accuracy alongside sophisticated performance. In this work we have tried to apply the various recent and state of the art actor-critic, policy gradient, both on-policy and off-policy Algorithms for the continuous control of the artificial cardiac pacemaker. We owe this work also due to the recent development of MATLAB® integration with the Reinforcement Learning toolbox in MATLAB® which combines low level RL algorithm tuning down to each and every hyperparameter and the high level model based control and electrical engineering tool that is Simulink®.