Abstract
Purpose:
To determine whether the performance of Argus II prosthesis users can be characterized by a limited set of visual function measures. Reducing the number of outcome measures needed to characterize performance would streamline the assessment of device benefit.
Methods:
The Argus II Retinal Prosthesis System includes a 10 x 6 electrode array implanted epiretinally, a tiny video camera mounted on a pair of glasses, and a small external computer that processes the video and determines the stimulation current of each electrode in real time. As part of a clinical trial, many outcome measure tests were repeatedly administered at consistent time points; performance was measured with the System ON and OFF. Additional outcome measures were performed only once on each subject. Here we determined how 14 different performance measures of Argus users correlate, including their ability to locate white objects on a black background, their ability to detect the direction of motion of an object, their spatial visual acuity, their ability to sort laundry based on gray scale, their ability to follow a line on the floor, their ability to find a door, and their ability to read printed letters. Most performance data were taken from the two-year follow up visit. A few tasks were only administered once during research projects. We then performed two types of multivariate analysis, Principal Component Analysis and Sparse Component Analysis, to study if the outcome measures can be reduced in dimensionality to three (locating objects, determining direction of motion and their spatial visual acuity) while adequately characterizing subject performance.
Results:
We find that the three measures (locating objects, determining direction of motion, and spatial visual acuity) correlate strongly with the other 11 measures. Both multivariate analysis measures reveal large orthogonal components between ON and OFF and between the three performance tasks; further dimensionality reduction results in significant loss of information.
Conclusions:
These results show that the overall performance of subjects is well described by the three outcome measures. The multivariate analysis results show that a complete description of subject performance needs to entail all three tasks. This suggests an overall paradigm for assessment of the performance of visual prosthesis users, which is significantly less burdensome than those employed previously.
Keywords: 607 nanotechnology •
754 visual acuity •
459 clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology