BCI + FES training for motor learning after stroke was feasible. A highly accurate brain signal control was achieved, and this signal could be reliably used to trigger the FES device for isolated index finger extension. With training, volitional control of isolated finger extension was attained in a small number of sessions. The source of motor recovery could be attributable to BCI, FES, combined BCI + FES, or whole arm or hand motor task practice.

Abstract-Twelve moderately to severely involved chronic stroke survivors (>12 mo) were randomized to one of two treatments: robotics and motor learning (ROB-ML) or functional neuromuscular stimulation and motor learning (FNS-ML). Treatment was 5 h/d, 5 d/wk for 12 wk. ROB-ML group had 1.5 h per session devoted to robotics shoulder and elbow (S/E) training. FNS-ML had 1.5 h per session devoted to functional neuromuscular stimulation (surface electrodes) for wrist and hand (W/H) flexors/ extensors. The primary outcome measure was the functional measure Arm Motor Ability Test (AMAT). Secondary measures were AMAT-S/E and AMAT-W/H, Fugl-Meyer (FM) upper-limb coordination, and the motor control measures of target accuracy (TA) and smoothness of movement (SM). ROB-ML produced significant gains in AMAT, AMAT-S/E, FM upper-limb coordination, TA, and SM. FNS-ML produced significant gains in AMAT-W/H and FM upper-limb coordination.

Recent neuroscience methods have provided the basis upon which to develop effective gait training methods for recovery of the coordinated components of gait after neural injury. We determined that there was not an existing observational measure that was, at once, adequately comprehensive, scored in an objectively-based manner, and capable of assessing incremental improvements in the coordinated components of gait. Therefore, the purpose of this work was to use content valid procedures in order to develop a relatively inexpensive, more comprehensive measure, scored with an objectively-based system, capable of incrementally scoring improvements in given items, and that was both reliable and capable of discriminating treatment response for those who had a stroke. Eight neurorehabilitation specialists developed criteria for the gait measure, item content, and scoring method. In subjects following stroke (>12 months), the new measure was tested for intra- and inter-rater reliability using the Intraclass Correlation Coefficient; capability to detect treatment response using Wilcoxon Signed Ranks Test; and discrimination between treatment groups, using the Plum Ordinal Regression. The Gait Assessment and Intervention Tool (G.A.I.T.) is a 31-item measure of the coordinated movement components of gait and associated gait deficits. It exhibited the following advantages: comprehensive, objective-based scoring method, incremental measurement of improvement within given items. The G.A.I.T. had good intra- and inter-rater reliability (ICC=.98, p=.0001, 95% CI=.95, .99; ICC=.83, p=.007, 95% CI=.32, .96, respectively. The inexperienced clinician who had training, had an inter-rater reliability with an experienced rater of ICC=.99 (p=.0001, CI=.97, .999). The G.A.I.T. detected improvement in response to gait training for two types of interventions: comprehensive gait training (z=-2.93, p=.003); and comprehensive gait training plus functional electrical stimulation (FES; z=-3.3, p=.001). The G.A.I.T. was capable of discriminating between two gait training interventions, showing an additive advantage of FES to otherwise comparable comprehensive gait training (parameter estimate=1.72, p=.021; CI, .25, 3.1).