ColourSpot, a novel gamified tablet-based test for accurate diagnosis of color vision deficiency in young children

Teresa Tang,John Maule,Leticia Álvaro,Anna Franklin,Alice Skelton,James Alvarez,Jenny Bosten

doi:10.3758/s13428-021-01622-5

Teresa Tang, John Maule + Show 5 more

Open Access

https://doi.org/10.3758/s13428-021-01622-5

Copy DOI

Abstract

There is a need for a straightforward, accessible and accurate pediatric test for color vision deficiency (CVD). We present and evaluate ColourSpot, a self-administered, gamified and color calibrated tablet-based app, which diagnoses CVD from age 4. Children tap colored targets with saturations that are altered adaptively along the three dichromatic confusion lines. Two cohorts (Total, N = 772; Discovery, N = 236; Validation, N = 536) of 4–7-year-old boys were screened using the Ishihara test for Unlettered Persons and the Neitz Test of Color Vision. ColourSpot was evaluated by testing any child who made an error on the Ishihara Unlettered test alongside a randomly selected control group who made no errors. Psychometric functions were fit to the data and “threshold ratios” were calculated as the ratio of tritan to protan or deutan thresholds. Based on the threshold ratios derived using an optimal fitting procedure that best categorized children in the discovery cohort, ColourSpot showed a sensitivity of 1.00 and a specificity of 0.97 for classifying CVD against the Ishihara Unlettered in the independent validation cohort. ColourSpot was also able to categorize individuals with ambiguous results on the Ishihara Unlettered. Compared to the Ishihara Unlettered, the Neitz Test generated an unacceptably high level of false positives. ColourSpot is an accurate test for CVD, which could be used by anyone to diagnose CVD in children from the start of their education. ColourSpot could also have a wider impact: its interface could be adapted for measuring other aspects of children’s visual performance.

Highlights

Measuring individual visual performance in young children has long been a challenge (Robbins et al, 2003)
We present results evaluating ColourSpot in a two-stage design that ensures that its validation is based on data from a cohort independent of the cohort used to identify optimal classification parameters for color vision deficiency (CVD) versus normal color vision
To find the optimal parameters of the model-free estimation of the psychometric function to achieve the best classification of CVD versus normal color vision for the discovery cohort, we made a systematic search of parameter space

Summary

Introduction

Measuring individual visual performance in young children has long been a challenge (Robbins et al, 2003). This has meant that visual problems like color vision deficiency. Anna Franklin and Jenny Bosten are joint senior authors (CVD) or other visual abnormalities are often not detected as early as they could be. There is a need to develop widely accessible intuitive psychophysical methods targeted to young children. One of the barriers to mass screening for CVD in children is a lack of widely accessible, high quality and child-appropriate diagnostic tests. Identifying CVD in young children is necessary to support them in accessing an educational system that relies heavily on color

Objectives

Methods

Results

Conclusion