October 17, 2022
By Arief Tjitra Salim, BEng, and Associate Professor Mohamed Dirani, PhD, MBA, GAICD
Given the predicted exponential increase in the incidence of myopia in the coming decades, having a tool that accurately predicts the visual function of any individual has great potential to assist not only the community, but also organizations at all levels and in all sectors of eye care to reduce the public health burden of myopia and its associated vision loss.
Myopia, or short-sightedness, is the most common cause of vision loss worldwide, affecting approximately one-third of the global population.1 Research has shown that the prevalence of myopia is increasing around the world, with up to half of the world’s population, or five billion people, predicted to have myopia by the year 2050.1 With the rising prevalence of myopia, there has also been a concomitant decrease in the average age at which children are developing myopia, resulting in faster rates of myopia progression.2 For this reason, by the time their myopia stabilizes during their teenage years or early adulthood, more people are suffering from a form of myopia commonly linked to sight-threatening conditions.3
As many as one billion people will have high myopia by the year 2050, placing a significant and potentially unmanageable proportion of the global population at risk of irreversible vision loss.1 Myopia and high myopia represent significant challenges to the global community in terms of personal and economic costs as well as loss of productivity. There is an urgent need to develop new interventions to manage this epidemic.
Current Treatments for Myopia: Limitations and the Need for Predictive Modeling
The management of myopia, including optical, therapeutic, and lifestyle interventions typically commence following diagnosis by a trained eye care professional through an eye examination. It has been reported that in some cases, by this time, the child’s myopia has already progressed, and their vision is affected. This is especially true for younger children who are unaware that their vision is abnormal. Given that more than 10% of children in some countries such as Singapore already have myopia before the age of 6 years, and that the average age of onset of myopia is decreasing, this could represent a substantial proportion of the population for whom vision is deteriorating undetected.4
Following the onset of myopia, an optometrist may commence a variety of treatments that correct the child’s myopia with spectacles or contact lenses and laser surgery in adulthood. In conjunction with treatment, control interventions to slow the progression of myopia, such as orthokeratology and therapeutics (atropine drops) are now being more widely adopted.5,6 Control interventions for myopia are essential because unlike simple optical corrections, they aim to slow the progression of myopia and hence reduce the risk of developing more sight-threatening forms of myopia.
Recent research has supported the idea that myopia may be prevented or slowed through non-medicinal interventions by managing its modifiable risk factors. A combination of outdoor activity and natural light exposure, looking far into the distance, taking regular breaks from near-work activities such as reading and smart devices, and undergoing annual eye check-ups during childhood have been associated with slowing or halting myopia onset and progression.7-9 However, as with many of the treatments mentioned above, these strategies are preventative and require early commencement to be most effective.
Initiating care-seeking and treatment is the responsibility of parents, and there is currently no way for parents to know their child’s risk of developing myopia reliably or accurately, when the condition may occur, to what extent, whether the condition may progress to high myopia, and what may be done to mitigate these risks. Strategies that utilize predictive modeling to determine an individual’s risk of developing myopia and determine the best management options would greatly benefit a wide range of stakeholders, including parents and eye health professionals.
Myopia Research: Where to Next?
Myopia has historically been under-prioritized in research, with greater emphasis placed on other eye diseases. This has occurred because myopia is so common and, in most cases, easily correctable, resulting in the perception that it is more of an optical inconvenience than a potentially serious health problem. However, in light of the alarmingly high prevalence of myopia globally and its profound economic costs ($244 billion annually in the U.S. in productivity losses alone),10 myopia has gained greater attention from academia and industry, and it is now a top health priority area for governments around the world. In fact, the Chinese government has even launched a work plan jointly released by eight national ministries, led by the Ministry of Education, that aims to reduce the rate of children’s myopia by 0.5% a year by 2030.
There is currently a big data revolution that, along with advances in artificial intelligence (AI), has begun to overcome previous limitations in data analytics and epidemiological research, and myopia research has begun to benefit from such innovation. The quantity of data available for analysis has increased exponentially in recent years, and data scientists can now link datasets, including clinical registries and population-based databases, providing an unprecedented wealth of information for big data analytics.
One Chinese study conducted in 2018 used refraction data, age, and myopia progression rates from electronic medical records in eight ophthalmic research centers to develop a machine learning algorithm to predict a child’s myopic status up to the age of 18 years.11 This study successfully predicted the extent of myopia in an individual up to eight years in advance with a moderate to high degree of accuracy, highlighting that AI and big data analytics can now be used to predict a child’s vision reliably.
This type of research holds great promise for future applications that allow personalized prediction of a child’s risk of myopia. Still, it is largely academic and conceptual and is not yet being applied in real-world settings (in the form of a user-facing platform) to predict the risk of developing myopia or high myopia in individual children. It is important to note that while this study generally made predictions of high accuracy, the accuracy diminished for predictions of refractive status in later years. One potential reason for inaccuracy may be the lack of real-time input data pertaining to individual risk factors, such as time spent outdoors, time spent on near tasks such as smart device use, and parental refractive status.
These limitations may be overcome, and the predictive accuracy of such algorithms strengthened using “digital phenotyping.” Digital phenotyping describes the real-time quantification of the individual-level human phenotype in situ using data from personal digital devices such as smartphones or computers.12 Data pertaining to many of the major risk factors for myopia can be passively monitored by mobile phones, especially given that the devices themselves are now considered a contributing risk factor.13 Digital phenotyping based on how people interact with their devices has already been shown to successfully predict depressive and psychotic episodes in psychiatric patients.14-16 This concept has been applied to ophthalmological research, including myopia, although to a limited extent.
One example of digital phenotyping in myopia research came from a 2021 study on 525 Dutch teenagers aged 12-16 years that investigated the link between smartphone use and refractive error. The study utilized a smartphone app (Myopia app; Innovattic) to gather objective data on screen time, face-to-screen distance, and the number of times of continuous smartphone use for more than 20 minutes.9 Similarly, researchers in Singapore developed a fitness tracker (FitSight) that aimed to capture objective data on outdoor time by estimating the real-time ambient light level of its user. The tracker utilized a companion smartphone app that records the patterns and provides feedback to their parents, with the goal of encouraging children to engage in more time outdoors.17
Introducing ‘Predict My Child’s Vision’ By Plano
There is a clear need for a personalized service based on real-time digital phenotyping, predictive analytics, and AI that can utilize individual-level risk factor data to predict long-term changes in a child’s vision accurately. This will allow parents to make informed decisions about how to intervene to slow or prevent myopia in their children. A Singapore-based health-tech company, Plano, has developed “Predict My Child’s Vision” to address this need.
Predict My Child’s Vision is an AI algorithm that utilizes machine learning and predictive analytics to reliably predict the likelihood that a child will develop myopia, at what age, to what extent, whether they will develop high myopia, and at what age their myopic deterioration is likely to stabilize. Based on the latest scientific evidence, the algorithm also provides parents with suggestions to manage their child’s individual risk factors so that they may maximize their chances of having healthy vision in the future.
Given the predicted exponential increase in the incidence of myopia in the coming decades, having a tool that accurately predicts the visual function of any individual has great potential to assist not only the community, but also organizations at all levels and in all sectors of eye care to reduce the public health burden of myopia and its associated vision loss.
Arief Tjitra Salim, BEng, is the Research and Operations Lead at Plano Pte Ltd.
Mohamed Dirani, PhD, MBA, GCAID, is the Founding Managing Director of Plano Pte Ltd and an Adjunct Associate Professor at the Duke-NUS Medical School. He is also an Adjunct Principal Investigator at the Singapore Eye Research Institute and an Honorary Principal Investigator at the Centre for Eye Research Australia.
References
1 B. A. Holden et al., “Global Prevalence of Myopia and High Myopia and Temporal Trends from 2000 through 2050,” (in eng), Ophthalmology, vol. 123, no. 5, pp. 1036-42, May 2016, doi: 10.1016/j.ophtha.2016.01.006.
2 I. G. Morgan et al., “The epidemics of myopia: Aetiology and prevention,” (in eng), Prog Retin Eye Res, vol. 62, pp. 134-149, Jan 2018, doi: 10.1016/j.preteyeres.2017.09.004.
3 S. Y. Chua et al., “Age of onset of myopia predicts risk of high myopia in later childhood in myopic Singapore children,” (in eng), Ophthalmic Physiol Opt, vol. 36, no. 4, pp. 388-94, Jul 2016, doi: 10.1111/opo.12305.
4 M. Dirani et al., “Prevalence of refractive error in Singaporean Chinese children: the strabismus, amblyopia, and refractive error in young Singaporean Children (STARS) study,” (in eng), Invest Ophthalmol Vis Sci, vol. 51, no. 3, pp. 1348-55, Mar 2010, doi: 10.1167/iovs.09-3587.
5 X. Yang, Z. Li, and J. Zeng, “A Review of the Potential Factors Influencing Myopia Progression in Children Using Orthokeratology,” (in eng), Asia Pac J Ophthalmol (Phila), vol. 5, no. 6, pp. 429-433, Nov/Dec 2016, doi: 10.1097/apo.0000000000000242.
6 Y. Y. Song, H. Wang, B. S. Wang, H. Qi, Z. X. Rong, and H. Z. Chen, “Atropine in ameliorating the progression of myopia in children with mild to moderate myopia: a meta-analysis of controlled clinical trials,” (in eng), J Ocul Pharmacol Ther, vol. 27, no. 4, pp. 361-8, Aug 2011, doi: 10.1089/jop.2011.0017.
7 J. C. Sherwin, M. H. Reacher, R. H. Keogh, A. P. Khawaja, D. A. Mackey, and P. J. Foster, “The association between time spent outdoors and myopia in children and adolescents: a systematic review and meta-analysis. [Review],” vol. 1, no. 10, pp. 2141-51, Oct 2012.
8 H. P. Board, “Health Promotion Board Annual Report 2009/2010,” Singapore, 2010. [Online]. Available: https://www.hpb.gov.sg/docs/default-source/annual-reports/hpb-annual-report-2010.pdf?sfvrsn=2
9 C. A. Enthoven et al., “Smartphone Use Associated with Refractive Error in Teenagers: The Myopia App Study,” (in eng), Ophthalmology, vol. 128, no. 12, pp. 1681-1688, Dec 2021, doi: 10.1016/j.ophtha.2021.06.016.
10 K. S. Naidoo et al., “Potential Lost Productivity Resulting from the Global Burden of Myopia: Systematic Review, Meta-analysis, and Modeling,” (in eng), Ophthalmology, Oct 17 2018, doi: 10.1016/j.ophtha.2018.10.029.
11 H. Lin et al., “Prediction of myopia development among Chinese school-aged children using refraction data from electronic medical records: A retrospective, multicentre machine learning study,” (in English), PLoS Medicine, vol. 15 (11) (no pagination), no. e1002674, November 2018, doi: http://dx.doi.org/10.1371/journal.pmed.1002674.
12 J. P. Onnela and S. L. Rauch, “Harnessing Smartphone-Based Digital Phenotyping to Enhance Behavioral and Mental Health,” (in eng), Neuropsychopharmacology, vol. 41, no. 7, pp. 1691-6, Jun 2016, doi: 10.1038/npp.2016.7.
13 H. Terasaki, T. Yamashita, N. Yoshihara, Y. Kii, and T. Sakamoto, “Association of lifestyle and body structure to ocular axial length in Japanese elementary school children,” (in eng), BMC Ophthalmol, vol. 17, no. 1, p. 123, Jul 12 2017, doi: 10.1186/s12886-017-0519-y.
14 S. Saeb, E. G. Lattie, S. M. Schueller, K. P. Kording, and D. C. Mohr, “The relationship between mobile phone location sensor data and depressive symptom severity,” (in eng), PeerJ, vol. 4, p. e2537, 2016, doi: 10.7717/peerj.2537.
15 G. Bedi et al., “Automated analysis of free speech predicts psychosis onset in high-risk youths.”
16 T. R. Insel, “Digital Phenotyping: Technology for a New Science of Behavior,” (in eng), Jama, vol. 318, no. 13, pp. 1215-1216, Oct 3 2017, doi: 10.1001/jama.2017.11295.
17 P. K. Verkicharla et al., “Development of the FitSight Fitness Tracker to Increase Time Outdoors to Prevent Myopia,” vol. 1, no. 3, p. 20, Jun 2017.
