How Digital Twins Could Transform Eye Care
A digital twin is a computer-based representation of part or all of the eye. It may combine imaging, clinical measurements and mathematical models to represent ocular structure, function or behaviour.
Researchers are investigating whether these models could eventually support disease monitoring, treatment research and surgical planning. However, most ophthalmic applications remain experimental, and many current studies involve virtual or reference models rather than complete digital twins that update continuously for an individual patient.
What Is a Digital Twin?
A digital twin is a computational representation of a physical structure or biological process that is linked to real-world data. In eye care, it might represent an individual eye, a specific tissue, visual function or a physiological process such as blood flow.
Definitions vary, but a complete clinical digital twin would generally be patient-specific, updated as new measurements become available and capable of producing predictions that can be compared with real outcomes. Many ophthalmic studies currently described as digital twins are virtual models, reference models or early components of this broader concept.
How Is It Different from Artificial Intelligence?
Artificial intelligence can recognise patterns or make predictions from data. A digital twin is a broader modelling concept that aims to represent how a particular structure or process behaves and may change under different conditions.
A digital twin may use AI, but it can also rely on mathematical equations, biomechanics, optical modelling or physiological simulations. An AI system that classifies an image is not automatically a digital twin.
Could Digital Twins Improve Diagnosis?
Digital twins might eventually help clinicians combine several measurements and compare changes over time. They could potentially highlight when an eye is behaving differently from its expected course.
This possibility has not yet been established for routine diagnosis. Any model would need to demonstrate that it improves clinical decisions when compared with established examinations, imaging and diagnostic pathways.
Could They Predict Disease Progression?
Repeated measurements could be used to update a model and estimate possible future changes. This could eventually help distinguish stable disease from conditions requiring closer review or treatment.
Most current studies are retrospective, cross-sectional or based on limited datasets. Prospective research is needed to show that digital-twin predictions accurately reflect what later happens to individual patients.
Personalised Treatment Planning
Researchers are developing digital models of the eye to simulate how different interventions might affect a model before they are used clinically. This may eventually help clinicians compare possible outcomes and risks. However, these tools have not been validated to select treatment for individual patients.
Key Points
| Aspect | Explanation | Current Status |
| Digital model | A computational model of an eye structure, function or biological process linked to clinical or experimental data | Definitions and technical maturity vary |
| Treatment simulation | Researchers can simulate how different interventions might affect the model | Not validated for routine treatment decisions |
| Potential benefit | May help compare possible outcomes and sources of risk | Promising but unproven |
| Limitations | Cannot reliably identify the best treatment for an individual patient | Requires prospective validation |
| Clinical use | Not established for routine patient-specific treatment selection | Must be compared with established care |
Cataract and Refractive Surgery
Patient-specific optical models can combine measurements such as corneal shape, axial length, anterior-chamber depth and estimated lens position. These methods may help researchers improve predictions of postoperative vision and lens performance.
Existing biometry, formulae and ray-tracing tools should not automatically be described as complete digital twins. A true digital-twin system would need to update with patient data and demonstrate reliable predictions across different eyes, procedures and clinical settings.
Corneal Conditions
Patient-specific computational models can simulate how the cornea deforms under pressure or after a procedure. Researchers are studying their potential use in keratoconus, corneal cross-linking, refractive surgery and selected corneal operations.
Most current models rely on assumptions about tissue properties that cannot be measured fully in an individual eye. Further validation is needed before they can routinely select treatment settings or predict postoperative stability.
Glaucoma
Digital-twin research in glaucoma includes models of optic-nerve biomechanics, retinal blood flow and the combined effects of intraocular and systemic blood pressure. These approaches aim to improve understanding of why eyes with similar pressure measurements may progress differently.
A 2026 physiology-based digital twin used individual intraocular-pressure and blood-pressure data to simulate ocular haemodynamics. It identified three haemodynamic profiles, one of which was associated with a higher risk of glaucoma progression, but the approach remains investigational and does not currently guide routine treatment.
Myopia
A 2026 study developed an OCT-based three-dimensional virtual whole-eye model known as CET-1. The study included 80 eyes from 40 participants across a wide range of axial lengths, while the direct comparison between CET-1 and MRI-derived models was reported for 70 eyes from 35 participants.
The researchers found close geometric agreement between the OCT- and MRI-derived models and demonstrated that CET-1 could quantify differences in ocular shape associated with myopia. However, the study did not prospectively test whether the model could predict future myopia progression, complications or treatment response in an individual patient.
Age-Related Macular Degeneration
Some research teams also use the term digital twin for data-driven reference models of cells or tissues rather than for a continuously updated model of an individual patient. In 2026, researchers created a detailed three-dimensional reference model of retinal pigment epithelial cell structure using imaging data from approximately 1.3 million laboratory-grown cells.
The model was designed to investigate how these cells develop their organisation and polarity, processes relevant to retinal disease research. It was not a patient-specific model and did not directly predict the development of age-related macular degeneration or an individual response to treatment.
Visual Function Testing
A 2024 study developed digital-twin models of contrast-sensitivity function using hierarchical Bayesian methods and historical testing data. When combined with new individual measurements, the models could estimate performance in untested conditions.
In one experimental setting using 25 trials, the researchers estimated that the approach could reduce data collection by more than 50%. This finding relates to research testing efficiency and does not mean that standard clinical vision assessments can currently be reduced by half.
Could They Improve Surgical Planning?
Patient-specific computational models may allow clinicians to explore how tissue position, shape or optical performance could change after a procedure. One early example is the Nebraska Nomogram for Autograft Planning, software developed to plan ipsilateral rotational autokeratoplasty using corneal imaging, thickness measurements and topography.
The published evaluation included retrospective testing on a previous case and clinical use in a further patient. This is preliminary evidence from a highly specialised procedure and does not show that digital-twin planning is ready for routine ophthalmic surgery or that simulated outcomes can guarantee safety, feasibility or postoperative vision.
Could They Support Remote Eye Care?
Home measurements such as visual test results, eye-pressure readings or retinal images could theoretically be used to update a digital model between appointments.
Reliable home devices, secure data transfer, quality control and clear clinical-response pathways would all be required. Continuously updated digital twins for remote ophthalmic management are not currently part of routine care.
Could Digital Twins Support Drug Research?
Cellular and tissue models may help researchers study biological pathways, generate hypotheses and prioritise treatments for laboratory testing. They could also support simulations of how a virtual population might respond differently to an intervention.
Digital twins cannot replace laboratory experiments or clinical trials. Predictions must be confirmed using biological models and appropriately designed human research before they can support treatment recommendations.
What Are the Main Technical Challenges?
Digital and AI-based eye models face several important technical challenges that can affect how reliable and useful they are in practice.
- Device variability: Eye data may come from different scanners and manufacturers, each using slightly different measurement, segmentation and calibration methods
- Generalisability: A model trained on one device, clinic or population may not perform as well in a different setting or patient group
- Biological complexity: Human biology is variable, so models need to reflect uncertainty rather than present a single prediction as definite
- Missing or incomplete data: Gaps in patient data can limit how accurately a model represents disease behaviour
- Changing disease patterns: Conditions can evolve over time, making it harder for models to stay accurate without regular updates
- Treatment adherence: Differences in how consistently treatments are followed can affect outcomes and reduce prediction reliability
- Measurement error: Small inaccuracies in imaging or data collection can influence model performance
To be clinically useful, these systems must be carefully validated across different settings and designed to handle uncertainty, rather than relying on a single fixed prediction.
Are There Privacy and Bias Concerns?
Digital twins may require detailed imaging, clinical histories and repeated measurements, making privacy, cybersecurity and appropriate access controls essential. Patients would need clear information about how their data are used, stored and shared.
Models must also be tested across different ages, ethnic groups, eye shapes and disease severities. Poorly representative datasets could produce less reliable predictions for groups that were under-represented during development.
Are Digital Twins Used Routinely Today?
Complete digital twins that continuously update and simulate the individual eye are not currently part of routine ophthalmic care. Existing research includes virtual anatomical models, physiological simulations, cellular reference models and patient-specific prediction tools.
These technologies require prospective clinical validation, comparison with established care and appropriate regulatory review. Current treatment decisions should continue to rely on recognised examinations, imaging, clinical guidelines and professional judgement.
Myth vs Fact
| Myth | Fact |
| Any three-dimensional eye image is a digital twin. | A digital twin requires a meaningful connection with real data and a model of structure, function or behaviour. |
| Every digital twin uses artificial intelligence. | Some rely mainly on mathematical, optical, physiological or biomechanical models. |
| AI image classification is the same as a digital twin. | Classification identifies patterns, while a twin aims to represent and simulate an individual system or process. |
| Digital twins can already predict which eye treatment will work best. | This remains a research goal and has not been established for routine care. |
| The 2026 myopia model predicted future progression. | It reconstructed and measured ocular shape but did not prospectively predict progression. |
| The NIH retinal-cell model represented an individual patient. | It was a laboratory reference model built from approximately 1.3 million cultured cells. |
| Digital twins can replace clinical trials. | Simulations must be confirmed through laboratory and human research. |
| Digital twins are already routinely used in UK eye clinics. | Complete patient-specific ophthalmic twins remain experimental. |
Key Takeaways
- Digital twins are computer models linked to real biological or clinical data.
- A digital twin may use mathematical, biomechanical, statistical or artificial-intelligence methods.
- Not every three-dimensional eye image or AI prediction model is a complete digital twin.
- Most ophthalmic applications remain at the research or early-development stage.
- Current studies include eye-shape reconstruction, glaucoma physiology and visual-function modelling.
- Some laboratory models described as digital twins represent cells or tissues rather than individual patients.
- Predictions require prospective and independent clinical validation.
- Digital twins do not currently replace eye examinations, OCT, visual-field testing or other established assessments.
- Clinical use would require appropriate data protection, cybersecurity and medical-device oversight.
Frequently Asked Questions
- What Is a Digital Twin in Eye Care?
A digital twin is a computational representation of an eye structure, visual function or biological process that is connected to real-world data. Complete patient-specific twins would be updated as new information becomes available. - How Is a Digital Twin Different from Artificial Intelligence?
AI identifies patterns or makes predictions from data. A digital twin aims to represent and simulate a structure or process and may use AI, mathematical equations, biomechanics or other modelling methods. - Could Digital Twins Improve Eye Disease Diagnosis?
Digital twins may eventually combine several measurements and identify changes that deserve closer assessment. They have not yet been validated as routine tools for diagnosing eye disease. - Can Digital Twins Predict Eye Disease Progression?
Potentially, but this has not been established for routine care. Prospective studies must show that predictions made by a digital twin accurately match later changes in individual patients. - Could Digital Twins Help Personalise Eye Treatments?
Researchers hope to simulate different treatment strategies using patient-specific measurements. Current evidence is not sufficient for digital twins to select treatment independently. - Could Digital Twins Improve Cataract or Refractive Surgery Planning?
A digital twin could combine information such as your corneal shape, lens position and eye length to create a detailed model of your eye. However, fully integrated digital twins are not currently used routinely for surgery planning. - Could Digital Twins Help Manage Glaucoma?
Glaucoma models are being developed to study optic-nerve biomechanics, blood flow and the interaction between eye pressure and blood pressure. They do not currently replace pressure measurements, OCT, visual fields or clinical assessment. - Could Digital Twins Support Remote Eye Monitoring?
Home-testing information could eventually update digital models between visits. Suitable devices, secure systems, quality checks and validated clinical-response pathways are still required. - What Are the Challenges of Using Digital Twins in Ophthalmology?
Challenges include collecting accurate data from different devices, managing missing information, protecting sensitive health data and ensuring predictions are reliable across different groups of people. - Are Digital Twins Used Routinely in Eye Care Today?
No. Ophthalmic research includes virtual anatomical models, cellular reference twins and experimental patient-specific simulations, but complete continuously updated digital twins are not currently used in routine eye care.
Final Thoughts: How Could Digital Twins Transform Eye Care?
Digital twins represent an emerging area of ophthalmic research. By combining clinical measurements with optical, biomechanical, physiological or data-driven models, they may eventually help researchers study disease progression, compare treatment strategies and improve planning for selected procedures.
Most current applications remain virtual models, reference twins or early patient-specific simulations rather than complete systems that update continuously and guide routine care. Prospective clinical validation, reliable data exchange, regulatory review and appropriate safeguards for privacy and bias will be needed before digital twins become part of everyday ophthalmology. To discuss concerns about your vision or eye health, contact Eye Clinic London to arrange an appropriate assessment.
References
- Drummond, D. and Gonsard, A. (2024) ‘Definitions and characteristics of patient digital twins being developed for clinical use: scoping review’, Journal of Medical Internet Research, 26, article e58504. Available at: https://pmc.ncbi.nlm.nih.gov/articles/PMC11602770/
- Katsoulakis, E., Wang, Q., Wu, H., Shahriyari, L., Fletcher, R., Liu, J., Achenie, L., Liu, H., Jackson, P., Xiao, Y., Syeda-Mahmood, T., Tuli, R. and Deng, J. (2024) ‘Digital twins for health: a scoping review’, npj Digital Medicine, 7(1), article 77. Available at: https://pmc.ncbi.nlm.nih.gov/articles/PMC10960047/
- Einighammer, J., Oltrup, T., Bende, T. and Jean, B. (2009) ‘The individual virtual eye: a computer model for advanced intraocular lens calculation’, Journal of Optometry, 2(2), pp. 70–82. Available at: https://www.sciencedirect.com/science/article/pii/S1888429609700275
- Xu, K., Chen, X., Liu, B., Jin, K., He, M. and Shi, D. (2025) ‘Digital twins in ophthalmology: concepts, applications, and challenges’, Asia-Pacific Journal of Ophthalmology, 14(6), article 100205. Available at: https://pubmed.ncbi.nlm.nih.gov/40378962/
- Zhao, Y., Lesmes, L.A., Dorr, M. and Lu, Z.-L. (2024) ‘Predicting contrast sensitivity functions with digital twins’, Scientific Reports, 14, article 24100. Available at: https://pubmed.ncbi.nlm.nih.gov/39406885/