Predicting task performance from upper extremity impairment measures after cervical spinal cord injury.

Abstract

Automated sensor-based assessments of upper extremity (UE) function after cervical spinal cord injury (SCI) could provide more detailed tracking of individual recovery profiles than is possible with existing assessments, and optimize the delivery and assessment of new interventions. The design of reliable automated assessments requires identifying the key variables that need to be measured to meaningfully quantify UE function. An unanswered question is to what extent measures of sensorimotor impairment can quantitatively predict performance on functional tasks. The objective was to define the predictive value of impairment measures for concurrent functional task performance in traumatic cervical SCI, as measured by the Graded Redefined Assessment of Strength, Sensibility and Prehension (GRASSP). Retrospective analysis. A data set of 138 GRASSP assessments was analyzed. The Strength and Sensation modules were used as measures of impairment, whereas the concurrent Prehension Performance module was used as the surrogate measure of function. Classifiers were trained to predict the scores on each of the six individual tasks in the Prehension Performance module. The six scores were added to obtain a total score. The Spearman's ρ between predicted and actual total Prehension Performance scores was 0.84. Predictions using both the Strength and Sensation scores were not found to be superior to predictions using the Strength scores alone. Measures of UE motor impairment are highly predictive of functional task performance after cervical SCI. Automated sensor-based assessments of UE motor function after SCI can rely on measuring only impairment and estimating functional performance accordingly.