Using EEG to Predict the Rehabilitation Outcome of Stroke Patients

Abstract

Interventions: Not Applicable. Main Outcome Measure(s): Performance-based clinical assessments of walking. Results: Item-bank development was accomplished using the first 2 qualitative stages of an 8-stage mixed-methods approach for assessment scale development (Velozo et. al, 2012). Stage 1 involved a comprehensive literature review to identify existing clinical assessment items that examine walking recovery. Walking recovery was defined as the ability to attain both steady-state walking and the ability to make adjustments to task and environmental demands (Patla and Shumway-Cook, 1998). A review of literature identified 15 existing clinical assessments that contained a subset of tasks assessing walking recovery. The total number of tasks identified was 138. Qualitative stage 2 involved eliminating redundant or misfit items (item winnowing) and item categorization based on dimensions of walking (item binning). The 138 tasks were reduced to 44 tasks, which were then grouped into 14 dimensions of walking recovery based on varying task demands (e.g., walking while carrying an object, walking and talking) or environmental demands (e.g., negotiating obstacles, varying terrain). Conclusions: A large bank of 44 items was identified from existing assessments that can comprehensively assess walking recovery post-stroke. Behavioral outcomes will be required to develop and validate a hierarchy of item difficulty in the item-bank. Computerized adaptive testing and item response theory based short forms could be used to reduce testing burden and enhance clinical utility of the item-bank for a comprehensive examination of walking recovery post-stroke.