Published: 2026-01-22 03:42
AI Offers Personalised Myopia Progression and Intervention Predictions
Myopia, commonly known as nearsightedness, is a global health challenge with rapidly increasing prevalence, particularly among children and young adults. Projections suggest that by 2050, half of the world’s population could be myopic, with a significant proportion developing high myopia, which carries substantial risks for sight-threatening complications.
In the UK, the rising incidence of myopia places a growing burden on ophthalmic services and public health. The need for more precise, individualised approaches to manage and slow its progression has become paramount.
Recent research, highlighted in *npj Digital Medicine*, points towards a transformative role for artificial intelligence (AI) in this field, offering personalised predictions for myopia progression and the efficacy of various interventions.
Understanding Myopia: A Growing UK Public Health Challenge
Myopia occurs when the eye grows too long from front to back, or the cornea is too steeply curved, causing light to focus in front of the retina rather than directly on it. This results in distant objects appearing blurred.
While often considered a simple refractive error corrected with spectacles or contact lenses, myopia, especially high myopia (typically defined as -6.00 dioptres or worse), is a significant risk factor for serious ocular pathologies.
Conditions such as retinal detachment, myopic maculopathy, glaucoma, and cataracts are more prevalent and can be more severe in highly myopic individuals. The economic and social impact of these conditions, including reduced quality of life and potential vision loss, underscores the urgency for effective management strategies.
The prevalence of myopia has been steadily increasing across the UK and globally. Lifestyle factors, such as increased near-work activities (e.g., screen time) and reduced outdoor time, are thought to contribute significantly to this trend.
Early onset and rapid progression of myopia in childhood are particularly concerning, as they often lead to higher levels of myopia in adulthood and a greater lifetime risk of associated complications.
Current Approaches to Myopia Management
Current clinical practice for managing myopia progression involves a combination of monitoring and intervention strategies. Regular eye examinations are crucial to track changes in refractive error and axial length (the length of the eyeball), which is a key indicator of myopia progression.
However, predicting the exact trajectory of an individual’s myopia progression remains challenging.
Monitoring Myopia Progression
- Refractive Error Measurement: Standard autorefraction and subjective refraction are used to determine the degree of myopia. Regular monitoring helps identify changes over time.
- Axial Length Measurement: Ocular biometry, often performed using optical coherence tomography (OCT) or other non-contact methods, provides precise measurements of the eye’s axial length. This is considered the most reliable indicator of myopia progression, as it directly reflects the underlying anatomical changes.
- Corneal Topography: Mapping the curvature of the cornea can be useful, especially when considering interventions like orthokeratology.
Intervention Strategies
Several evidence-based interventions are available to slow myopia progression, aiming to reduce the final degree of myopia and thus mitigate the risk of associated complications. These interventions are typically categorised as optical, pharmacological, or environmental.
Optical Interventions
- Myopia Control Spectacle Lenses: These lenses incorporate specific designs, such as peripheral defocus technology, to slow axial elongation.
- Multifocal Contact Lenses: Specialised soft contact lenses designed with different powers in the centre and periphery to create myopic defocus on the peripheral retina, which is thought to inhibit eye growth.
- Orthokeratology (Ortho-K): Rigid gas-permeable contact lenses worn overnight to temporarily reshape the cornea. While providing clear vision during the day without correction, they also induce peripheral myopic defocus, which can slow myopia progression.
Pharmacological Interventions
Low-dose atropine eye drops have emerged as a significant pharmacological option for myopia control. Atropine, an anticholinergic agent, is thought to slow myopia progression by acting on muscarinic receptors in the retina and sclera, influencing eye growth.
Clinical studies have demonstrated its efficacy in reducing the rate of progression, particularly at very low concentrations, which minimises side effects such as glare and reduced accommodation.
It is crucial for clinicians to understand the evidence base, potential side effects, and appropriate patient selection when considering pharmacological interventions. The decision to prescribe any medicine must always be based on a thorough clinical assessment and in accordance with relevant professional guidelines and regulatory frameworks.
Environmental and Lifestyle Modifications
- Increased Outdoor Time: Spending more time outdoors, particularly in natural light, has been consistently linked to a reduced risk of myopia onset and progression.
- Reduced Near-Work Strain: Encouraging regular breaks during close-up activities (e.g., the “20-20-20 rule”: every 20 minutes, look at something 20 feet away for 20 seconds) and maintaining appropriate working distances can help.
- Ergonomics: Ensuring good posture and lighting during reading or screen use.
Despite these available strategies, a significant challenge remains: how to accurately predict which intervention will be most effective for a particular individual. Current clinical decisions often rely on population-level data and a degree of trial-and-error, leading to varied outcomes and potentially delayed optimal management.
The Promise of AI in Myopia Prediction
The advent of artificial intelligence, particularly machine learning, offers a paradigm shift in how myopia progression might be predicted and managed. AI models have the capacity to analyse vast, complex datasets, identifying intricate patterns and correlations that are often beyond human cognitive ability.
This capability is particularly valuable in a multifactorial condition like myopia, where progression is influenced by a combination of genetic, environmental, and lifestyle factors.
The core promise of AI in this context lies in its ability to deliver truly personalised predictions. Instead of relying on average progression rates derived from large cohorts, an AI model can consider an individual patient’s unique profile – their age, ethnicity, parental myopia history, baseline refractive error, axial length, previous progression rates, and even lifestyle habits – to generate a tailored prognosis.
Such a personalised approach could revolutionise clinical decision-making. Clinicians could move from reactive management to proactive, data-driven intervention.
The research highlighted in *npj Digital Medicine* underscores this potential, suggesting that AI can guide predictions not only on how myopia will progress but also on the likely effectiveness of specific interventions for that individual.
How AI Models Might Work in Myopia Management
An AI system designed for myopia prediction would typically involve several stages, from data collection and model training to prediction generation and clinical integration.
Data Input and Feature Extraction
The effectiveness of any AI model hinges on the quality and quantity of the data it is trained on. For myopia prediction, this would involve a comprehensive array of clinical and demographic information:
- Demographic Data: Age at onset, current age, ethnicity, gender.
- Family History: Parental myopia status (presence, severity).
- Baseline Ocular Parameters: Initial refractive error (spherical equivalent), axial length, corneal curvature.
- Progression History: Rates of change in refractive error and axial length from previous examinations.
- Lifestyle Factors: Time spent outdoors, near-work duration, screen time (though these can be challenging to quantify accurately).
- Genetic Markers: While still largely research-based, incorporating genetic predisposition data could further enhance predictive power in the future.
These diverse data points, or “features,” are fed into machine learning algorithms. The algorithms then learn to identify complex relationships between these features and the ultimate outcome (myopia progression or response to intervention).
Machine Learning Techniques
Various machine learning techniques could be employed, each with its strengths:
- Neural Networks (Deep Learning): Particularly effective at identifying complex, non-linear patterns in large datasets, making them suitable for multifactorial conditions.
- Random Forests and Gradient Boosting Machines: Ensemble methods that combine multiple decision trees to improve accuracy and robustness, often providing insights into feature importance.
- Support Vector Machines: Algorithms that find the optimal hyperplane to classify data points, useful for distinguishing between different progression trajectories.
The AI model would be trained on historical patient data where both progression and intervention outcomes are known. Through this training, the model learns to associate specific input features with particular outcomes, allowing it to make predictions on new, unseen patient data.
Output and Clinical Utility
The output from an AI model could be multifaceted, providing clinicians with actionable insights:
- Personalised Risk Scores: A quantitative assessment of an individual’s likelihood of rapid myopia progression over a defined period (e.g., 1, 2, or 5 years).
- Predicted Progression Curves: Visualisations showing the anticipated trajectory of refractive error and axial length change, allowing for clearer communication with patients and parents.
- Intervention Efficacy Predictions: Estimates of how effective different myopia control strategies (e.g., specific multifocal contact lenses, orthokeratology, low-dose atropine) might be for that particular patient, expressed as a percentage reduction in progression or a predicted final refractive error.
- Optimal Intervention Pathways: Recommendations for the most suitable intervention or combination of interventions based on the patient’s profile and predicted response.
This level of detailed, personalised information would empower clinicians to make more informed decisions, tailor treatment plans, and set realistic expectations with patients and their families.
Clinical Implications for UK Practice
The integration of AI-guided predictions into UK ophthalmic and optometric practice holds significant potential to enhance patient care, optimise resource allocation, and improve long-term visual outcomes.
Enhanced Early Detection and Risk Stratification
AI models could assist in identifying children at high risk of developing or rapidly progressing myopia even before significant changes are clinically evident. By analysing baseline data and family history, AI could flag individuals who warrant closer monitoring or earlier intervention.
This proactive approach could prevent children from reaching high levels of myopia, thereby reducing their lifetime risk of severe ocular complications.
Optimised Intervention Selection
One of the most impactful applications of AI would be in guiding intervention choices. Instead of a trial-and-error approach, clinicians could use AI predictions to select the most effective myopia control strategy for an individual.
For example, if an AI model predicts that a particular child would respond significantly better to orthokeratology than to multifocal contact lenses, this information could guide the initial treatment decision, saving time and potentially achieving better outcomes.
This data-driven approach could lead to more efficient use of resources, reducing the need for multiple treatment changes and associated follow-up appointments.
Improved Patient and Parent Engagement
Visualising predicted progression curves and understanding the likely efficacy of different interventions can significantly improve communication with patients and parents. When presented with clear, data-backed prognoses, families are often better equipped to understand the importance of adherence to treatment and lifestyle modifications.
This enhanced engagement can lead to better compliance and, consequently, more successful myopia control.
Resource Optimisation within the NHS
The UK’s National Health Service (NHS) faces constant pressure to manage resources efficiently. AI tools could help optimise clinic workflows by identifying patients who require intensive management versus those who can be monitored less frequently.
This could free up valuable clinician time and clinic capacity, allowing for more targeted care where it is most needed. For instance, low-risk patients might be managed with less frequent follow-ups, while high-risk individuals receive more intensive, personalised care.
Integration with Existing Systems
For AI tools to be truly effective in UK practice, they must seamlessly integrate with existing electronic health record (EHR) systems and diagnostic equipment. This would allow for automated data input and output, minimising administrative burden and ensuring that AI predictions are readily accessible at the point of care
Source: Nature
