Information transmission into the human nervous system is the basis for a variety
of prosthetic applications. Spinal cord stimulation (SCS) systems are widely available,
have a well documented safety record, can be implanted minimally invasively, and are
known to stimulate afferent pathways. Nonetheless, SCS devices are not yet used for
computer-brain-interfacing applications.Here we aimed to establish computer-to-brain
communication via medical SCS implants in a group of 20 individuals who had been operated
for the treatment of chronic neuropathic pain.In the initial phase, we conducted interface
calibration with the aim of determining personalized stimulation settings that yielded
distinct and reproducible sensations. These settings were subsequently utilized to
generate inputs for a range of behavioral tasks. We evaluated the required calibration
time, task training duration, and the subsequent performance in each task.We could
establish a stable spinal computer-brain interface in 18 of the 20 participants. Each
of the 18 then performed one or more of the following tasks: A rhythm-discrimination
task (n = 13), a Morse-decoding task (n = 3), and/or two different balance/body-posture
tasks (n = 18; n = 5). The median calibration time was 79 min. The median training
time for learning to use the interface in a subsequent task was 1:40 min. In each
task, every participant demonstrated successful performance, surpassing chance levels.The
results constitute the first proof-of-concept of a general purpose computer-brain
interface paradigm that could be deployed on present-day medical SCS platforms.
