Ml4san

ML4SAN (Machine Learning for Safe and Accurate Neuroprostheses) Brain-to-body interfaces (BBIs) have been used to restore hand movements in patients with paralysis or stroke, but current technology only allows for limited and coarse movements. To improve daily functioning, new neuroprostheses are needed that can control individual fingers, provide precise grasping movements, and decode multiple degrees of freedom without causing rapid fatigue. This project aims to establish research collaborations focused on the development of novel decoding and control algorithms for the stimulation of neural or muscular structures to restore safe and reliable hand functionalities. These efforts align with my expertise and long-term research agenda in applying signal processing and machine learning to neuroprosthetic applications, and build upon my existing funding history and research vision for developing innovative neurotechnologies. To achieve this goal, I will undertake research visits to two international centres of excellence, namely the group of Prof. Silvestro Micera at the École Polytechnique Fédérale de Lausanne (Switzerland) and the group of Prof. Mike I. Jordan at INRIA Paris (France). Prof. Micera is a leader in the field of implantable neural interfaces, robotic systems for neurorehabilitation, and decoding and control algorithms. His group has made significant contributions to the development of bi-directional hand prostheses and other neuroprosthetic systems and is currently working on a neuroprosthetic system to restore grasping in paralysed individuals. This proposed research collaboration will build upon and complement their current research efforts. By working together, I can benefit from their expertise and existing network of collaborators and clinical partners while also bringing my own perspectives and insights. Prof. Jordan is a world-leading researcher in statistical machine learning with four decades of experience bridging the gap between machine learning and other fields. He has made major contributions in probabilistic graphical models, optimisation techniques, variational inference, and to our understanding of optimality principles for human motor control. His research groups at UC Berkeley and INRIA Paris work on game theory, causal inference, reinforcement learning, and decision-making under uncertainty. Within this project, Prof. Jordan's expertise will be instrumental in developing more intelligent and adaptive neuroprostheses that effectively balance performance with probabilistic safety. While the visits will be focused on completing our identified project, I also intend to use this opportunity to engage in wider discussions with the two research groups about areas of mutual interest, such as robotic rehabilitation, neural decoding, statistics and machine learning. The purpose of these research visits is to stimulate new ideas and collaborations and to generate preliminary results to support the development of grant proposals for follow-on collaborative research. This project also aims to promote and strengthen international connections between UK researchers and international leaders in their respective fields.