Rapid Neural Typing Interface for Veterans with ALS

Intracortical brain-computer interfaces (iBCIs) record and decode neural signals to enable fast, accurate and intuitive control of assistive technologies for Veterans and others living with paralysis arising from spinal cord injury, stroke, or amyotrophic lateral sclerosis (ALS). Using an intracortical BCI, people with tetraplegia have been able to use their imagined hand and finger movements to command point-and-click actions on a computer, type with a virtual keyboard, use communication apps such as chat, and browse the web. Recently, a person with tetraplegia used the investigational BrainGate iBCI, and specifically a decoder that could detect the attempted, individual movements of all 10 digits, to play songs (sequences of individual notes) using an on-screen piano keyboard. Remarkably, imagined left and right hand and digit movements could be decoded with high accuracy from neural signals recorded from a single hemisphere of motor cortex. Building on these observations, this project aims to develop a novel, intuitive neural typing interface for high-throughput communication by Veterans with ALS. The research hypothesizes and evaluates several methods by which digit and hand decoding could be combined to achieve direct neural control of a multi-row on-screen keyboard and simultaneous trackpad actions that would dramatically increase communication rates for individuals with paralysis. To further enhance neural decoding of imagined simultaneous bilateral hand and digit movements, this project will leverage motor cortical signals recorded from both hemispheres. Achieving effective and reliable neural decoding of intended, dexterous finger movements will require expanding current understanding of the encoding principles of neuronal firing and field potentials from both hemispheres – separately and in interaction – and creating new areal interaction models for the control of volitional bimanual actions. This project will assess the relationship between neural encoding of discrete digit/hand actions associated with imagined typing and encoding of point-and-click actions associated with continuous on-screen mouse control. To facilitate rapid switching between continuous mouse actions and rapid neural typing, we will seek latent spaces that characterize these actions in motor cortical activity and identify approaches to maximally disambiguate these two modes for optimal real- time decoding. This research leverages state-of-the-art machine learning innovations toward a more capable, reliable, and versatile iBCI to promote independence for Veterans and others with severe motor and motor impairments.