Implantable and wearable medical devices are increasingly being developed as alternative therapies for intractable diseases. In particular, undertreated neurological disorders such as epilepsy, migraine, and Alzheimer’s disease are of major public health concern around the world, driving the need to explore such new approaches. Despite significant advances in neural interface systems, the small number of recording channels in existing technology remains a barrier to their therapeutic potential. This is mainly due to the fact that simultaneous recording from a large number of electrodes imposes stringent energy and area constraints on the integrated circuits that interface with these electrodes. In this talk, I will first discuss an efficient compressive sensing framework for multichannel cortical implants. Next, I will present the design of our sub-microwatt per channel closed-loop seizure control device and both its in-vivo and offline performance. I will then discuss our latest work on the integration of machine learning algorithms for on-chip classification of neural data. Finally, I will give examples of how these results may be used towards designing new devices, to enhance the lives of millions of people suffering from disabling neurological conditions in future.