| dc.description.abstract | Purpose: Brain-computer interface (BCI) techniques may provide a link between an individual’s neurological activity and communication device control, which circumvents the requirement for individuals to possess a reliable form of physical movement for augmentative and alternative communication (AAC) device access. However, while BCI technology is rapidly progressing in the laboratory setting, BCI developments are advancing largely without consideration of established AAC best practices, which are crucial for effective clinical implementation of BCI technology. For instance, BCI research largely utilize custom made software and display paradigms and view BCI as a ‘one size fits all’ solution. That BCI is a one size fits all solution contrasts with AAC best practice, which seek to pair an individual to an AAC device that matches their current and future profile, communication needs, and preferences. Therefore, to bring BCI research further in line with existing AAC best practices this dissertation work aims to evaluate initial and recurring person-centered factors associated with learning of motor execution-based BCI switch for accessing a commercial AAC row-column scanning paradigm. Method: Four individuals with a diagnosis of amyotrophic lateral sclerosis (ALS) completed 12 BCI training sessions in which they made letter selections during an automatic row-column scanning pattern from a 7x5 grid. Neural signals utilized for BCI selection control were generated by motor execution during target letter highlighting. For comparison, three individuals without neurological impairment completed three BCI training sessions. During each session, participants completed approximately 20 minutes of online BCI. To assess person-centered factors associated with BCI performance and longitudinal device learning, participants completed both initial and recurring assessment measures. Initial assessment measures of an individual’s unique profile prior to BCI training included evaluation of neural signals utilized for BCI control (i.e., maximum event related synchronization amplitude (ERS), maximum event related synchronization amplitude minus predicted noise floor, and event related synchronization minus desynchronization difference; ERS-ERD), along with screening of cognitive factors, physical motor abilities, and motor imagery skills via the ALS-Cognitive Behavioral Screen, BCI screener (Pitt & Brumberg, 2018b), ALS-Functional Rating Scale, Bimanual Fine Motor Function, and Manual Ability Classification System. Recurring measures were taken during each BCI training session to evaluate changes associated with longitudinal BCI performance, and included measures of fatigue, motivation, time since last meal, device satisfaction, level of frustration with device control, mental and physical effort, and overall ease of device control. Results: Three out of four participants demonstrated either BCI performance in the range of neurotypical peers, or an improving BCI learning trajectory across sessions. However, while BCI learning trajectories for row-column scanning BCI device were variable both between and within participants for those with ALS, findings indicate that approximately five sessions were needed to generally characterize an individual’s learning trajectory during motor execution-based BCI trials. Regarding participant profiles, cognitive screening revealed that the two participants presenting with a suspicion for cognitive impairment achieved the highest levels of BCI accuracy, with their increased levels of performance being possibly supported by largely unimpaired motor skills. In addition, while scores for the cognitive section of the BCI screener were high, the two participants who did not demonstrate a consistent learning trajectory each missed one point in the area of attention and working memory, and one point in the area of cognitive motor learning and abstract problem solving. As expected, prior to BCI use, the greatest amplitude for each neurophysiological measure was generally associated with the highest levels of BCI accuracy. However, this finding was not consistent across sessions as the participant demonstrating the lowest amplitudes prior to BCI performance presented with the highest amplitudes during BCI control. Furthermore, when evaluating neurophysiological measures across sessions, a significant correlation between left hand peak ERS and BCI performance was identified for one participant. Finally, ERS-ERD measure remained highest for the participant achieving the highest level of BCI accuracy and was significantly correlated to BCI performance for the participant achieving the second highest BCI performance levels. For recurring number scale-based recurring measures: 1) ratings of motivation were high for all participants with ALS. However, motivation ratings significantly decreased across sessions for two participants, 2) while satisfaction ratings were positively correlated to BCI performance for two participants, satisfaction ratings for the other two participants were primarily driven by perceived levels of frustration, and 3) mental effort ratings significantly decreased across sessions for one participant along with improved BCI performance, and overall mental effort ratings showed a moderate negative trend with BCI performance for two participants. Conclusion: Overall findings support that (motor) imagery-based BCI switch access to a commercial AAC row-column scanning paradigm may be feasible for individuals with ALS, and that clinical decisions regarding BCI suitability may be informed through approximately 5 BCI training sessions, when using motor execution as a BCI control strategy. Furthermore, while generalization of findings is limited due to the small sample size, results provide multiple directions to help facilitate BCI’s clinical transition by informing BCI assessment and intervention procedures. Regarding BCI assessment, findings provide early guidelines governing the length of device trials for BCI paradigms based on motor execution, and support 1) ideally beginning BCI intervention before severe deterioration of physical motor abilities to facilitate BCI access across the disease course, facilitate BCI success, and support those with cognitive impairments, 2) further research into the development of BCI specific assessment tools, including neurophysiological measures of ERS and ERS-ERD difference to help standardize procedures for identifying factors related to BCI control. Findings relevant to BCI intervention include 1) incorporation of communication tasks beyond copy spelling to support sustained levels of BCI motivation, 2) incorporating a range of recurring person-centered measures in evaluating BCI trial outcomes including performance accuracy, levels of satisfaction, multiple measures of fatigue, and levels of frustration due to potentially differing definitions of fatigue, and differences in factors driving levels of BCI satisfaction 3) supporting more natural levels of mental effort during the establishment of BCI control. | |
