Bioelectric signals govern the functionality of a number of vital human organs, like the brain, heart, muscles, gut, etc. Since bioelectric anomalies can hint to severe health conditions, acting on the basis of this data is key for smart diagnostics and real time therapy. The state-of-the-art approach uses bio-compatible sensor and actuator arrays integrated in an organ conformal substrate, which enables local data gathering and actuation. However, no existing system has embedded computing capability for local interpretation of the data and millisecond decision-making, as required for real-time life-saving therapy. This project proposes a novel computing solution that can be reliably and monolithically integrated with existing sensor and actuators platforms. The proposed technology is based on an emerging resistive switching device called memristor (also known as RRAM). Metal-oxide memristors are super-dense, nonvolatile memory devices and hence ideal candidates for implementing energy-efficient switches and artificial synapses useful for biomedical applications. In this system, the memristors will be utilized in two complementary ways. Three-terminal biocompatible memristive switches will be developed to hide unwanted sensors and actuators (for example sensors that fall on nerve tissue or blood vessels) from the array. Memristors will be also utilized as non-volatile analog synapses for the implementation of a hardware neural network for the local analysis of the bioelectric data. Our cross-disciplinary effort will have a quantum leap effect on the development of an integratable hardware ANN that can compute data in the vicinity of the sensors and actuators. Such an integrated system capable of millisecond response times will enable groundbreaking technologies, like ultra-fast detection and therapy for heart diseases.