 Imagine if the Bluetooth on your phone could connect to your brain. There's been a great buzz about this topic lately, brain machine interfaces. You may have even heard the likes of Elon Musk with his new initiative to merge the human brain with artificial intelligence in order to save humanity in the future. There are, however, currently real problems too. People that are completely paralyzed or locked in. If this technology could communicate just one command per second, it could be used to control a wheelchair. Or for communication, on average we type it around 40 words per minute. So what can current technology achieve? Wearable technology, like EEG for example, uses electrodes placed externally on the surface of the scalp. These systems can only achieve a few characters per minute. This is because the electrodes are relatively far away from the brain and so they observe the activity of many millions of neurons. And so this is very noisy and for this reason implantable systems are being considered. A good example here is the BrainGate system developed in the US. This uses a tiny microelectroderay implanted in the motor cortex to record activity, decode it and control a cursor. Although there has been great progress here, there are some challenges. Firstly, the information transfer rate is still quite low. So far in this talk I've spoken around 200 words. It would take the best technology currently about half an hour to communicate the same message. Secondly, there is variability in the recordings, meaning a significant amount of setup time is required every day to get going. And thirdly, there are still wires coming through the skin. And so this is very much the focus of my research group at Imperial College London. How can technology address these grand challenges and what will implants of the future look like? So a good starting point here is to go back and look what is successful currently? Cochlear implants. And so these devices are typically implanted underneath the skin behind the ear. They have a small coil for receiving power, a metal package for protecting the electronics, and an array of electrodes for interfacing to the tissue. The most advanced systems have around 20 electrodes needing 20 separate wires. But brain machine interfaces will need hundreds, preferably thousands of electrodes. We need more channels to improve this information transfer rate, but also to have redundancy so it's stable day in, day out. As we are unable to increase the number of wires in these implants, we need a different approach. And this is what I'm working on, a next generation implantable technology. By miniaturizing the implant by about a thousand times using semiconductor technology. All aspects of the implants are being integrated into a tiny chip about the size of a grain of rice. One implant alone will observe the activity from a few tens of electrodes. But then by using multiple implants, we can achieve the scalability we're required to hundreds or thousands. But for this to work, each device needs to be independent, autonomous and fully wireless. So to get the power in and the data out, we are developing a two-tier network. This will have wireless links across the skin and across the dura. That's the body's air to blood and blood to brain barrier. In addition to achieving improved efficiency, this will also reduce the risk of infection. There is however also another problem of power, that any power the implant needs to operate will be dissipated as heat. And more than a one degree C increase in temperature will actually damage the tissue. So if we need more channels, the power per channel needs to reduce. Fortunately, this is an area my group has some expertise in, in low power chip design. We achieve this by adopting a bio-inspired holistic approach. Firstly, by drawing inspiration from biology, in the way the electronics represent and process data, we can massively improve efficiency. Secondly, by considering the system as a whole, we can make big savings compared to optimizing each component individually. We can do this in this case because the chip itself is the whole system. Using this approach, we've already demonstrated a 10 times reduction in both power and area of brain machine interfaces. So when we put all this together, we will have a next generation implantable technology. It will allow us to effectively and reliably communicate between a living brain and our IT systems. So the question I leave you with is not if, but when will all this be available? Thank you.