Are we equipped to utilize AI and other digital advancements for effective myopia management?
Driven largely by factors such as decreased outdoor time and increased near-work activities, the prevalence of myopia is predicted to affect half of the world’s population by 2050. Various interventions and the potential role of digital technologies and artificial intelligence in myopia management are significant topics of discussion in this field. However, despite promising advancements, challenges persist in effectively integrating these technologies into clinical practice.
Currently, childhood myopia management comprises various interventions, including environmental interventions such as increasing time outdoors, optical interventions like peripheral myopic defocus spectacles and orthokeratology, and pharmaceutical interventions such as atropine eye drops. Monitoring and providing ongoing care for this growing number of patient populations is a tremendous task for eye health professionals.
While leveraging innovative digital technologies and harnessing the power of artificial intelligence (AI) may offer some solutions for myopia detection and management, the question remains: Are we there yet?
Challenges in integrating AI into myopia care
Dr. Monica Chaudhry, Director of Monica Chaudhry Vision Institute (MCVI), commented on the emergence of AI in various aspects of eye care. “Although artificial intelligence and myopia control are still in their early stages, there is significant development happening worldwide at the moment,” she said.
Despite AI seemingly infiltrating many aspects of our lives, its applications in myopia management may not yet be ready for primetime.
As we’ve seen in other areas of ophthalmology, there is potential for AI to streamline early identification, risk stratification, progression prediction, and optimization of timely interventions.
Dr. Chaudhry believes that AI could play a role in diagnosing and managing patients with myopia. “With the help of AI, myopia progression can be estimated, and this plays a significant role in better understanding its progression and, hence, better management or improved strategy planning for individual patients,” she added.
However, it is important to note that the performance of any AI model is limited by the quality and quantity of its reference datasets, which may include both clinical data and imaging
Building the foundation: The importance of large datasets
In a study conducted by Zhang and Zou,1 the current application status of AI in myopia was reviewed, focusing on the data modalities used for developing AI models. They found that even though studies have reported effective implementation of AI in clinical practice, there are critical technical and clinical restrictions that must be overcome.
They highlighted the complex multifactorial aspects of myopia development and progression, emphasizing the necessity for large, high-quality public datasets to improve the current capabilities of handling multimodal input.
Dr. Jeffrey J. Walline, a distinguished professor and acting dean at The Ohio State University College of Optometry, is also carefully watching emerging AI and digital technologies with cautious optimism. Within his practice, he sees first-hand the growing need for management strategies. “I’m sure there is a use for artificial intelligence in myopia management, but I don’t know what it is yet,” he shared.
As highlighted in the review by Zhang and Zou, the multifactorial nature of myopia and its clinical course can make the application of a single algorithm very challenging, and Dr. Walline agrees. “I think myopia control is a long, long way from personalized treatment because it is so complex. We have individualized treatments for cancer because some forms of cancer are caused by single gene mutations. However, more than 100 loci have been identified as related to myopia. When we combine all of those loci, we still explain very little of the variance in refractive error,” he explained.
In order to address all the variabilities in clinical and genetic presentation, the reference databases for AI and machine learning would need to be broad and extremely large to deliver the sensitivity required to translate into clinical acceptance and uptake.
Deep learning for myopia detection
Deep learning, a subset of the AI branch of machine learning, can automatically extract rules from a reference or known dataset to make judgments about new or unknown data. Multiple studies on deep learning for detecting various eye conditions have shown incredible accuracy, in some cases outperforming trained personnel and clinicians.
Since 2018, several research groups have evaluated whether deep learning can be used to accurately evaluate refractive error from imaging. By evaluating features of retinal photographs or other ocular appearance images, trained deep learning systems have demonstrated the ability to detect refractive error and myopia with a high degree of accuracy. This type of application can be highly impactful for the assessment of children, nonverbal patients, and patients in remote areas using telemedicine.
One branch of AI-driven myopia detection has demonstrated excellent performance in recent years. Pathologic myopia is unique in that it is associated with excessive axial elongation, leading to structural changes in the posterior segment, such as posterior staphyloma and myopic macular degeneration, ultimately resulting in the loss of visual acuity.
A review of 14 AI-based models to detect pathologic myopia and associated complications, based on fundus and OCT imaging, demonstrated excellent performance.2
Results based on imaging are indeed interesting and promising. However, variations in procedures, databases, imaging samples, and other aspects of methodology make integration into our daily practice challenging.
Dr. Walline has also been carefully watching the work being done with remote image analysis, and he is hopeful that this may translate into more accurate prediction models with clinical applications.
“We may see advances in predicting who is more likely to experience sightthreatening complications. The use of fundus (retinal) photographs and/ or OCT images to predict ocular and systemic complications has rapidly progressed with artificial intelligence. So perhaps that will be the next area of discovery at the crossroads between myopia and technology,” enthused Dr. Walline.
Genetics-x-Environment: Go outside and play!
A 2024 publication from Biswas et al. revealed that myopia-related genetics have remained relatively stable and that this surge in myopia can be linked to a complex interplay in factors, with strong associations shown between decreased outdoor time in childhood and an increase in excessive engagement in near work.3
So, what is it about being outside that prevents myopia? It is thought that the protective effect against myopia may primarily be attributed to retinal dopamine released with exposure to natural light, such as sunlight. Dopamine has been proposed to influence eye growth and play a role in emmetropization. When released by the amacrine cells of the retina, it has shown a dose-response relationship with the intensity of light exposure.
It would be short-sighted not to recognize that myopia is no longer an emerging significant global public health problem with serious socioeconomic consequences: It is already here!
If you aren’t myopic yourself, you likely know someone who is. In recent decades, we have witnessed a concerning increase in the prevalence of myopia (spherical equivalent of -0.50 diopters and below), and even high myopia (spherical equivalent of -5.0 diopters and below).
Currently, it is estimated that approximately one in three people globally have myopia, and over 399 million have high myopia. A metaanalysis of 145 studies published by Holden and colleagues predicted that by 2050, half of the world’s population (49.8%) will be living with myopia, and nearly one in 10 (9.8%) will have high myopia.4
The rise in myopia rates is predominantly observed in East Asian urban populations, where up to 90% of young adults are myopic (with 20% classified as highly myopic), and up to 62% of 12-year-old children exhibit some degree of myopia.5,6 Additionally, countries that previously had relatively lower rates of myopia have recently shown a doubling of rates compared to three decades ago.7
Due to its childhood onset and the importance of maintaining visual function and quality of life, early screening for myopia, timely diagnosis, and therapeutic interventions are crucial.
Digital devices targeting myopia prevention
As we noted, the rise in myopia may be associated with less time spent outdoors in childhood and our obsession with devices and other near-work activities. Perhaps a digital technology that doesn’t require prolonged staring could be the solution to modifying detrimental behaviors from an early age.
Vivior Monitor is a wearable device created in Switzerland to evaluate visual behavior in children with myopia between 6 and 16 years old. They found that as children got older, they spent less time looking at distant objects and engaging in physical activity, opting instead to spend more time on computers or similar devices.
The confirmation that outdoor activity has a protective effect against myopia has led to the development of wearable technology aimed at encouraging children to spend more time outside.
Sony has developed watch-based trackers that can provide feedback on the amount of time spent on indoor and outdoor activities and recommend ways to improve. Other sensor devices, such as the Clouclip, can be attached to the individual’s glasses and alert them of risky visual behaviors, such as prolonged nearwork, poor lighting, or short viewing distances.
With that being said, digital technology might have a role to play in treating myopia.
MyopiaX, a digital treatment currently undergoing clinical trials in Europe, is intended to slow the progression of myopia in children and adolescents. This smartphone app, used with a virtual reality headset and a Bluetooth controller, delivers a light stimulus during each ‘play’ session on top of the virtual reality game. The purpose is to modulate retinal dopamine levels, affecting the progression of myopia.
Another digital therapeutic for pediatric myopia is S-Alpha Therapeutic’s SAT-001, aimed at delaying pediatric myopia progression by modulating the balance of neuronal-humoral factors using a proprietary software algorithm. It provides instructions to perform specific tasks, such as eye exercises and low-intensity daily exercises, to satisfy the physical parameters. On the other hand, the emotional parameter guides the patient to feel relaxed, comfortable, safe, satisfied, entertained, and accomplished.
Balancing innovation with patient care
As we’ve seen with other technological advances in different branches of eye care, their integration into our daily practices requires addressing a number of challenges. These include the usual issues associated with ethics, legality, and regulation. However, even if this new application, whether digital or AI-related, is ‘interesting’ or ‘cool’, the bottom line is that it has to improve patient care delivery while lessening the burden on clinicians, the care team, and patients.
There is incredible potential for what AI models can achieve in terms of individualized treatment and precision medicine for myopia, but only if they have access to big medical data.
Dr. Chaudhry welcomes AI and digital technology into the myopia care space, stating, “Prevention is key and AI will help empower individuals and eye care practitioners to control myopia, which is expected to impact half of our population by 2050.”
References
- Zhang J, Zou H. Insights into artificial intelligence in myopia management: from a data perspective. Graefes Arch Clin Exp Ophthalmol. 2024;262(1):3-17.
- Zhang Y, Li Y, Liu J, Wang J, Li H, Zhang J, Yu X. Performances of artificial intelligence in detecting pathologic myopia: a systematic review and meta-analysis. Eye (Lond). 2023;37(17):3565-3573.
- Biswas S, El Kareh A, Qureshi M, Lee DMX, Sun CH, Lam JSH, Saw SM, Najjar RP. The influence of the environment and lifestyle on myopia. J Physiol Anthropol. 2024;43(1):7.
- Holden BA, Fricke TR, Wilson DA, Jong M, Naidoo KS, Sankaridurg P, et al. Global prevalence of myopia and high myopia and temporal trends from 2000 through 2050. Ophthalmology. 2016;123(5):1036-1042.
- Jung SK, Lee JH, Kakizaki H, Jee D. Prevalence of myopia and its association with body stature and educational level in 19-yearold male conscripts in Seoul, South Korea. Invest Ophthalmol Vis Sci. 2012;53(9):5579- 5583.
- Ding BY, Shih YF, Lin LLK, Hsiao CK, Wang IJ. Myopia among school children in East Asia and Singapore. Surv Ophthalmol. 2017;62(5):677- 697.
- Vitale S, Sperduto RD, Ferris FL, 3rd. Increased prevalence of myopia in the United States between 1971- 1972 and 1999-2004. Archi Ophthalmol. 2009;127:1632-1639.
Editor’s Note: This article was published in COOKIE Magazine Issue 15.