Methods and systems using fractional rank precision and mean average precision as test-retest reliability measures

Abstract

Disclosed herein are methods and systems of evaluating test-retest precision using fractional rank precision or mean-average precision, comprising: a) collecting a test and a retest result of each subject, wherein the results are described in feature space(s) and collected from a vision test machine; b) selecting, a first test result of a first subject; c) calculating distances from the first test result to the retest result of each subject; d) assessing, a similarity between the first test result and the retest result of each subject by ranking the distances in a non-descending order; e) assessing a rank precision for the first subject based on a rank of a distance from the first test result to the retest result of the first subject; f) repeating b), c), d), and e) for each subject; and evaluating, the test-retest precision based on the rank precision for each of the plurality of subjects.

What is claimed is: 
     
         1 . A computer-implemented method of evaluating test-retest precision of a test and a retest of a plurality of subjects using fractional rank precision (FRP) or mean-average precision (MAP), comprising:
 a) collecting, by a computer, a test result and a retest result of each of the plurality of subjects, wherein the test result and the retest result are described in one or more feature spaces and one of the test result and the retest result is collected from a vision test machine;   b) selecting, by the computer, a first subject from the plurality of subjects and a first test result of the first subject;   c) calculating, by the computer, distances from the first test result to the retest result of each of the plurality of subjects;   d) assessing, by the computer, a similarity between the first test result and the retest result of each of the plurality of subjects by ranking the distances in a non-descending order;   e) assessing, by the computer, a rank precision for the first subject based on a rank of a distance from the first test result to the retest result of the first subject;   f) repeating, by the computer, steps b), c), d), and e) for each of the plurality of subjects; and   g) evaluating, by the computer, the test-retest precision based on the rank precision for each of the plurality of subjects.