Development of a system based on artificial intelligence to identify visual problems in children: study protocol of the TrackAI project

Validation of a New Digital and Automated Color Perception Test

Although color vision deficiencies are very prevalent, there are no ideal methods for assessing color vision in all environments. We compared a new digital and automated method that quantifies color perception for the three protan, deutan, and tritan axes with two of the most commonly used color tests in daily practice: the Ishihara 38 plates test and the Farnsworth-Munsell 100-Hue test. One hundred patients underwent a triple examination composed of the new DIVE Color Test, the Ishihara test, and the Farnsworth-Munsell 100-Hue test. The DIVE Color Test was performed twice in forty participants to assess its repeatability. In the trichromatic group, the mean age stood at 20.57 ± 9.22 years compared with 25.99 ± 15.86 years in the dyschromatic group. The DIVE and Ishihara tests exhibited excellent agreement in identifying participants with color deficiency (Cohen's kappa = 1.00), while it was 0.81 when comparing DIVE and Farnsworth. The correlation between the global perception values of Farnsworth (TES) and DIVE (GCS) was 0.80. The repeatability of the DIVE Color Test was high according to Bland-Altman analysis with an intraclass correlation coefficient of 0.83. According to Ishihara, the DIVE Color Test proved to be an effective and reproducible tool for red-green color vision deficiency detection, capable of determining the severity of the defect in each of the three axes faster and more accurately than both Ishihara and Farnsworth.

Colour perception develops throughout childhood with increased risk of deficiencies in children born prematurely

AIM

To quantify the impact of prematurity on chromatic discrimination throughout childhood, from 2 to 15 years of age.

METHODS

We recruited two cohorts of children, as part of the TrackAI Project, an international project with seven different study sites: a control group of full-term children with normal visual development and a group of children born prematurely. All children underwent a complete ophthalmological exam and an assessment of colour discrimination along the three colour axes: deutan, protan and trytan using a DIVE device with eye tracking technology.

RESULTS

We enrolled a total of 1872 children (928 females and 944 males) with a mean age of 6.64 years. Out of them, 374 were children born prematurely and 1498 were full-term controls. Using data from all the children born at term, reference normative curves were plotted for colour discrimination in every colour axis. Pre-term children presented worse colour discrimination than full-term in the three colour axes (p < 0.001). Even after removing from the comparison, all pre-term children with any visual disorder colour discrimination outcomes remained significantly worse than those from full-term children.

CONCLUSION

While colour perception develops throughout the first years of life, children born pre-term face an increased risk for colour vision deficiencies.

Artificial intelligence-based screening for amblyopia and its risk factors: comparison with four classic stereovision tests

Introduction

The development of costs-effective and sensitive screening solutions to prevent amblyopia and identify its risk factors (strabismus, refractive problems or mixed) is a significant priority of pediatric ophthalmology. The main objective of our study was to compare the classification performance of various vision screening tests, including classic, stereoacuity-based tests (Lang II, TNO, Stereo Fly, and Frisby), and non-stereoacuity-based, low-density static, dynamic, and noisy anaglyphic random dot stereograms. We determined whether the combination of non-stereoacuity-based tests integrated in the simplest artificial intelligence (AI) model could be an alternative method for vision screening.

Methods

Our study, conducted in Spain and Hungary, is a non-experimental, cross-sectional diagnostic test assessment focused on pediatric eye conditions. Using convenience sampling, we enrolled 423 children aged 3.6-14 years, diagnosed with amblyopia, strabismus, or refractive errors, and compared them to age-matched emmetropic controls. Comprehensive pediatric ophthalmologic examinations ascertained diagnoses. Participants used filter glasses for stereovision tests and red-green goggles for an AI-based test over their prescribed glasses. Sensitivity, specificity, and the area under the ROC curve (AUC) were our metrics, with sensitivity being the primary endpoint. AUCs were analyzed using DeLong's method, and binary classifications (pathologic vs. normal) were evaluated using McNemar's matched pair and Fisher's nonparametric tests.

Results

Four non-overlapping groups were studied: (1) amblyopia (n = 46), (2) amblyogenic (n = 55), (3) non-amblyogenic (n = 128), and (4) emmetropic (n = 194), and a fifth group that was a combination of the amblyopia and amblyogenic groups. Based on AUCs, the AI combination of non-stereoacuity-based tests showed significantly better performance 0.908, 95% CI: (0.829-0.958) for detecting amblyopia and its risk factors than most classical tests: Lang II: 0.704, (0.648-0.755), Stereo Fly: 0.780, (0.714-0.837), Frisby: 0.754 (0.688-0.812), p < 0.02, n = 91, DeLong's method). At the optimum ROC point, McNemar's test indicated significantly higher sensitivity in accord with AUCs. Moreover, the AI solution had significantly higher sensitivity than TNO (p = 0.046, N = 134, Fisher's test), as well, while the specificity did not differ.

Discussion

The combination of multiple tests utilizing anaglyphic random dot stereograms with varying parameters (density, noise, dynamism) in AI leads to the most advanced and sensitive screening test for identifying amblyopia and amblyogenic conditions compared to all the other tests studied.

Distributed Intelligent Learning and Decision Model Based on Logic Predictive Control

By the method of documentation and logical analysis, based on the data, based on logic and based on the knowledge of three kinds of artificial intelligence in the sports education, the intelligent learning system feedback delay are studied, combined with mobile communication which led to the artificial intelligence online sports games teaching, pattern recognition, and virtual technology combined with innovative teaching interaction and experience. Promoting the development of green PE teaching machine learning can identify the types of PE activities and realize efficient PE learning diagnosis. Intelligent decision support system can identify sports talents and improve the effect of personalized PE teaching evaluation. From the perspective of psychological development and education, the key problems to be solved in the integration of artificial intelligence and physical education are examined. Then, the consistent model predictive control for feedback delay of nonlinear sports learning multiagent system with network induced delay and random communication protocol is studied. Under the communication waiting mechanism designed, each agent has a certain tolerance of delay, and this tolerance can be determined by ensuring the stability of the system. At the same time, a random communication protocol is designed to ensure the ordered communication of the multiagent system. Finally, the effectiveness of the proposed algorithm is verified by numerical simulation. To solve the channel competition access problem of the sports intelligent learning system with special structure feedback delay model predictive control, a dual channel awareness scheduling strategy under the model predictive control framework was proposed, and the distributed threshold strategy of sensors and the priority threshold strategy of controllers were designed. It is proved that the sensor will eventually work at Nash equilibrium point under the policy updating mechanism, and the priority threshold strategy of the controller is better than the traditional independent and identically distributed access strategy. By avoiding the data transmission when the channel status is poor, the channel access of the system is efficient and saves energy.

Recent Advances in Screening and Treatment for Amblyopia

Amblyopia is the most common cause of monocular visual impairment in children, with a prevalence of 2-3%. Not only is visual acuity reduced in one eye but binocular vision is affected, fellow eye deficits may be present, eye-hand coordination and reading can be affected, and self-perception may be diminished. New technologies for preschool vision screening hold promise for accessible, early, and accurate detection of amblyopia. Together with recent advances in our theoretical understanding of amblyopia and technological advances in amblyopia treatment, we anticipate improved visual outcomes for children affected by this very common eye condition. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Machine Learning Applications in Pediatric Ophthalmology

Purpose: To describe emerging applications of machine learning (ML) in pediatric ophthalmology with an emphasis on the diagnosis and treatment of disorders affecting visual development. Methods: Literature review of studies applying ML algorithms to problems in pediatric ophthalmology. Results: At present, the ML literature emphasizes applications in retinopathy of prematurity. However, there are increasing efforts to apply ML techniques in the diagnosis of amblyogenic conditions such as pediatric cataracts, strabismus, and high refractive error. Conclusions: A greater understanding of the principles governing ML will enable pediatric eye care providers to apply the methodology to unexplored challenges within the subspecialty.