Decoding Motor Signals From the Pediatric Cortex: Implications for Brain-Computer Interfaces in Children

Abstract

To demonstrate the decodable nature of pediatric brain signals for the purpose of neuroprosthetic control. We hypothesized that children would achieve levels of brain-derived computer control comparable to performance previously reported for adults. Six pediatric patients with intractable epilepsy who were invasively monitored underwent screening for electrocortical control signals associated with specific motor or phoneme articulation tasks. Subsequently, patients received visual feedback as they used these associated electrocortical signals to direct one dimensional cursor movement to a target on a screen. All patients achieved accuracies between 70% and 99% within 9 minutes of training using the same screened motor and articulation tasks. Two subjects went on to achieve maximum accuracies of 73% and 100% using imagined actions alone. Average mean and maximum performance for the 6 pediatric patients was comparable to that of 5 adults. The mean accuracy of the pediatric group was 81% (95% confidence interval [CI]: 71.5-90.5) over a mean training time of 11.6 minutes, whereas the adult group had a mean accuracy of 72% (95% CI: 61.2-84.3) over a mean training time of 12.5 minutes. Maximum performance was also similar between the pediatric and adult groups (89.6% [95% CI: 83-96.3] and 88.5% [95% CI: 77.1-99.8], respectively). Similarly to adult brain signals, pediatric brain signals can be decoded and used for BCI operation. Therefore, BCI systems developed for adults likely hold similar promise for children with motor disabilities.