To report on the utility of a computer tablet–based method for automated testing of visual acuity in children based on the principles of game design. We describe the testing procedure and present repeatability as well as agreement of the score with accepted visual acuity measures.
Reliability and validity study.
Setting : Manchester Royal Eye Hospital Pediatric Ophthalmology Outpatients Department. Patient Population : Total of 112 sequentially recruited patients. Intervention : For each patient 1 eye was tested with the Mobile Assessment of Vision by intERactIve Computer for Children (MAVERIC-C) system, consisting of a software application running on a computer tablet, housed in a bespoke viewing chamber. The application elicited touch screen responses using a game design to encourage compliance and automatically acquire visual acuity scores of participating patients. Acuity was then assessed by an examiner with a standard chart-based near ETDRS acuity test before the MAVERIC-C assessment was repeated. Main Outcome Measure : Reliability of MAVERIC-C near visual acuity score and agreement of MAVERIC-C score with near ETDRS chart for visual acuity.
Altogether, 106 children (95%) completed the MAVERIC-C system without assistance. The vision scores demonstrated satisfactory reliability, with test-retest VA scores having a mean difference of 0.001 (SD ±0.136) and limits of agreement of 2 SD (LOA) of ±0.267. Comparison with the near EDTRS chart showed agreement with a mean difference of −0.0879 (±0.106) with LOA of ±0.208.
This study demonstrates promising utility for software using a game design to enable automated testing of acuity in children with ophthalmic disease in an objective and accurate manner.
Amblyopia is the most common cause of visual impairment in children (prevalence in childhood of 1%–4%) and the leading cause of monocular vision loss in the 20- to 70-year- old age group. There are many other significant causes of central visual loss in children, including cataract, corneal opacity, and retinal disease. Crucial to the management of these conditions is an accurate and reliable assessment of vision, appropriate to the precise clinical needs. This may range from testing of a broad number of detailed visual functions by specialists in hospital patients to more basic vision screening of schoolchildren in vans. Recent attention has been directed to computerized vision testing, and many applications are readily available for desktop, laptop, mobile, and tablet devices, with several publications addressing this potential. However, most of these computerized systems are for testing adults and are not validated for automatic testing of vision in children without expert assistance.
This paper presents assessment of a system of automated vision testing in children using a customized computer tablet–based acuity test. Its key features are that it is housed in its own controlled viewing environment and uses game design principles to automatically present appropriate graphical targets, eliciting responses from children without the need for external intervention. It has been developed through an extensive, iterative period of testing and redesign and is based on an adult test (Mobile Assessment of Vision by intERactIve Computer; MAVERIC) on which we have previously reported.
We conducted a reliability and validity study to assess a novel system of automated vision testing in children (Mobile Assessment of Vision by intERactIve Computer for Children; MAVERIC-C) in terms of its reliability and agreement with standard methods of vision assessment, as well as its acceptability to pediatric hospital patients. The research adhered to the tenets of the Declaration of Helsinki and ethics committee approval for the full testing protocol was obtained prior to the start of the study through the U.K. independent research approval system. Informed consent was obtained for all patients, who were recruited from a general pediatric ophthalmology clinic in a teaching hospital in England.
The structure and development of the fundamental aspects of the MAVERIC system have been previously described in detail in its application to adults. In essence, it consists of a computer tablet loaded with specifically designed software housed in a custom-made viewing booth. The software generates a diminishing-sized square array resembling a Landolt “C” in the center of the screen and requires the user to detect the location of the gap. The subject responds by pressing 1 of the 4 surrounding buttons that corresponds to the gap location. The button provides audible and visible feedback and the software is programmed to give verbal encouragement to the subjects if they fail to respond to a target within a time limit. The game design element of the software is implemented by custom-made graphics and specific animations when the correct or incorrect response is given. For the system used in testing children (MAVERIC-C) the animation graphics and game design were enhanced to make the process more appealing to children. The onscreen graphics present either a mouse going into a hole, a pig into a pen, or a sheep into a pen when the correct responses are given, with multiple additional animations and sounds to encourage a child’s participation. The MAVERIC vision testing strategy involves 4 phases. Phase 1 involves a screening test to derive an initial rapid and approximate threshold. Phase 2 involves detailed threshold detection; 3 out of 4 correct responses are required to progress to the next reduced target size. If this is failed, the same target size is repeated; and if failed twice in a row, the phase is ended and the threshold taken as the last sequence of 3 correct responses out of 4. By the end of this challenging stage children’s concentration might be waning, so an additional simple suprathreshold test with new graphics is incorporated as phase 3 before a final repeated detailed threshold-level test. This principle of using multiple tests concurs with other established vision-testing algorithms.
The tablet used in the current study was the Galaxy Tab Pro 8.4 SMT 320 (Samsung Electronics, South Korea), selected for its high screen resolution. The display resolution of 1600 × 2560 pixels over an 8.4-inch-diagonal screen size allowed for the smallest letter size to be −0.21 logMAR (Snellen equivalent 6/3.7 or 20/12.3) at the testing distance of 40 cm. The next step was 0.09 logMAR (6/7.4 or 20/24.7), as this was the next possible size according to the screen resolution. For this study the maximum testing size used was 1.22 logMAR (6/100 or 20/333.3) and the device was calibrated with a photometer such that the central target luminance was black (0.57 cd/m 2 ) while the surrounding luminance was set to the maximum of 397.6 cd/m 2 . Overall contrast was therefore 99%.
One hundred and twelve children were enrolled for the study from a typical pediatric ophthalmology outpatient clinic. Patients were excluded only if they had a physical disability that excluded use of a tablet computer. To incorporate a significant range of visual acuities into the study, the eye chosen for the trial was the eye with worst visual acuity, up to 1.22 logMAR (6/100). If the vision was the same in both eyes, the right eye was chosen. All children wore habitual correction for all vision testing, with additional correction for 40 cm near testing if they were pseudophakic.
Before a computerized visual assessment was conducted, children were shown the tablet computer outside its booth and given a few minutes to familiarize themselves with the MAVERIC-C game. When the patient showed that he or she understood and could perform the basic test, the tablet was placed inside the booth and the patient invited to look through the viewing aperture, placing his or her hands inside the booth to provide a comfortable location from which to operate the tablet. For all children the fellow eye was occluded using a patch. The patient began the visual testing by pressing a large central green “start” button on the tablet’s screen. Once the test started, no further external input was given while the test ran through levels automatically, giving programmed audio encouragement where required. When the final acuity was determined a cheer sounded to signify the end of the examination. Masked to the MAVERIC-C vision result, the examiner then tested the near visual acuity of the patient using an ETDRS logMAR chart (Precision Vision, Lasalle, United States, SKU 2112). The examiner positioned the near visual acuity chart at the 40 cm test distance from the bridge of the child’s nose using the attached cord. The acuity score was recorded as a logarithm of the minimum angle of resolution (logMAR) value for the last line in which the subject identified 3 or more optotypes on that line, plus a value of −0.02 log unit for each optotype that was identified correctly beyond that line.
Finally, approximately 20 minutes after the first test, a second MAVERIC-C test was initiated.
The test-retest reliability of the MAVERIC-C system was assessed using the Bland-Altman limits of agreement (LOA) method to assess agreement between 2 measures. We used the same procedure to assess the agreement of the MAVERIC-C score with the standard near acuity charts.
A total of 112 children (52 male) were entered into the study, aged between 4 and 16 years (mean: 10.2 ± 2.82 years). Of the 112 subjects recruited, 106 were able to complete the test without any further assistance (95%). Testing algorithms dictated that the time to completion of the whole test from pressing start to the finishing cheers was between 3 and 6 minutes and average time was approximately 5 minutes. Those who could not complete the test had difficulty with understanding or willingness to play and were excluded from further analysis.
The range of pathologies of the children included 45 with primary diagnoses of anterior segment disorders (27 with keratoconjunctivitis, 14 with uveitis, 4 with cataract). There were 27 children with strabismus and amblyopia, including 11 with convergent squint (4 alternating eso) and 6 with divergent squint (1 intermittent). There were 14 children entered into the study with oculoplastic disorders and 16 with neuroophthalmology disorders, including 4 with nystagmus. There were 10 children with other miscellaneous diagnoses or for whom there was no abnormality found. The distribution of near visual acuity (displayed in Figure 1 ) was fairly broad.