BioAge Scientific Evidence and Clinical Interpretation

1. Why Biological Age Matters in Preventive Care

Chronological age is an important predictor of health, but it does not fully explain why people of the same age may have markedly different health spans. Biological age attempts to characterize this individual variation by estimating physiological aging, which comprises the cumulative effects of time, genetics, environment, and modifiable health factors on the body.

Measures of biological age have been associated with morbidity and mortality beyond chronological age alone (1,2). These biological “clocks”—derived from molecular, physiological, and imaging data—can estimate aspects of aging at the whole body, organ, and cellular level and are being investigated for earlier risk identification and prevention (3). Among the contributors to biological aging are modifiable factors such as blood pressure, metabolic health, physical activity, sleep, smoking, and diet (3,4). Biological age may therefore capture the cumulative effects of risk factors that clinicians assess and manage individually.

2. The Retina as a Source of Systemic Health Information

The retina provides a unique, clinically accessible in vivo view of neural tissue and microvasculature, with well-established anatomical, physiological, and epidemiological links to systemic health (5,6). Retinal images can therefore reveal information beyond ocular health. Fundus photography provides high-resolution visualization of retinal vessels, often capturing features that are difficult to observe through a routine eye exam. These vessels are exposed to many of the same systemic stressors that affect microcirculation elsewhere in the body, including elevated blood pressure, hyperglycemia, inflammation, and aging. Resulting vascular changes may therefore reflect broader microvascular health (7).

Decades of large population-based epidemiological research have linked quantifiable retinal vascular changes, including arteriolar narrowing and venular widening, with subsequent cardiovascular and cerebrovascular outcomes. The Atherosclerosis Risk in Communities (ARIC) study demonstrated associations between retinal vessel caliber and long-term cardiovascular outcomes (8). The Blue Mountains Eye Study and subsequent meta-analyses similarly linked retinal vascular features with incident stroke and coronary heart disease, often independent of traditional risk factors (9,10). Hypertensive retinal changes have also long been recognized as manifestations of hypertensive target-organ damage (11) and included in hypertension guidelines (12).

Artificial intelligence (AI), especially deep learning, has made retinal image analysis more efficient and consistent. Historically, subtle vascular features often required specialist grading and centralized reading centers. AI models can now identify these complex patterns, including some not readily recognized through conventional human assessment. This contributed to the emergence of oculomics, the study of systemic health information derived from ocular data (13). In 2018, a deep-learning model estimated age, sex, smoking status, systolic blood pressure, and cardiovascular risk from fundus photographs (14). A subsequent multicohort study extended this work to detect additional systemic biomarkers and risk factors (15), and systematic reviews have summarized the growing field of AI-based retinal cardiovascular-risk assessment (16). While most clinical evidence remains observational or retrospective, it supports the premise that retinal images contain measurable information related to vascular and metabolic health.

Prospective studies testing the clinical utility of AI analysis of retinal images for systemic health assessment are emerging. With the millions of retinal images already being acquired as part of care, AI analysis may provide additional systemic health information without a blood draw or laboratory testing. Together, biological plausibility, epidemiological evidence, advances in image analysis, and clinical studies provide the foundation for retinal biological-age measures.

3. Understanding BioAge

Interpreting BioAge requires a clear understanding of what it represents, its intended role, and its limitations.

3.1 What BioAge measures

BioAge is a cardiovascular-focused biological age estimate, expressed in years, derived from retinal images using an underlying cardiovascular-risk model. This deep-learning model analyzes a standard fundus photograph together with a limited set of demographic and health inputs to estimate a Pooled Cohort Equation–derived 10-year atherosclerotic cardiovascular disease (ASCVD) risk score (17). BioAge then uses peer-matching and regression methods to translate this cardiovascular-risk information into an age-equivalent estimate (18). Put simply, it asks: how old are the people whose retinal-based cardiovascular-risk profiles most closely resemble this individual’s?

The BioAge model was developed as an extension of this retinal ASCVD-risk model and has been evaluated using UK Biobank and EyePACS data (18). The resulting BioAge is therefore an age-equivalent estimate based on a cardiovascular-risk construct, rather than a direct measurement of blood pressure, blood glucose, cholesterol, or any single clinical variable.

3.2 The BioAge gap

A BioAge result has limited meaning without comparison to chronological age. The BioAge gap compares the retinal aging pattern with chronological age and is calculated as BioAge minus chronological age. For example, a BioAge of 58 years in a 55-year-old individual produces a gap of +3 years. A gap above zero reflects a retinal aging pattern more typical of older peers, while a gap below zero reflects a pattern more typical of younger peers. A gap near zero is broadly aligned with the individual’s chronological age.

A single result does not show whether biological aging is improving or worsening over time, and it should not be treated as a precise physiological measurement, diagnostic threshold, or treatment threshold. Differences of several years should be interpreted proportionately and alongside established clinical information.

3.3 Scientific basis and factors associated with the BioAge gap

BioAge is grounded in the field of oculomics, which examines how ocular data relates to systemic health and aging. Direct studies have linked BioAge with cardiovascular-kidney-metabolic (CKM) health measures and telomere length, while independent retinal-age models have been associated with aging-related outcomes (18–21). Though these findings support biological plausibility, they do not establish prospective clinical utility. Section 5 reviews the evidence in detail.

Biological and cardiovascular aging reflect an individual’s genetics, environment, lifestyle, and long-term exposure to cardiometabolic risk factors. In one research analysis, BioAge was related to an American Heart Association-based cardiovascular health score (20). This analysis also found that blood pressure was the most important contributor to variation in the BioAge gap, although the relative contributions of individual factors require further study. This relationship is plausible because sustained blood-pressure elevation can alter retinal vessel structure and caliber, reflecting current and prior exposure (22). Retinal analysis may therefore capture aspects of cumulative vascular exposure not represented by a single measurement. BioAge does not identify the cause of an individual result; a BioAge gap above zero may reflect multiple influences and should not be attributed to any single risk factor.

3.4 Clinical role and interpretive boundaries

Based on current evidence, BioAge is best framed as complementary information that may support patient engagement and preventive-health conversations, alongside established assessment. BioAge application will depend on the clinical setting and scope of practice. In eye care, BioAge may reinforce the importance of systemic health assessment. In primary, preventive, concierge, or longevity-focused care, it may be considered alongside blood pressure, lipid and glucose measures, medical history, and guideline-based CKM health assessment.

BioAge does not replace guideline-based risk assessment and clinical evaluation. A favorable result does not make routine preventive care unnecessary. BioAge is not a diagnostic test and does not diagnose cardiovascular disease, hypertension, diabetes, kidney disease, or any other condition.

4. BioAge Within the Biological Age Landscape

No universally accepted gold standard exists for biological age (1), and an age expressed in years does not make these various measures equivalent. Biological-age measures draw on diverse inputs, including DNA methylation, blood biomarkers, proteomic profiles, medical images, wearable sensors, and health records. The models also differ in their training targets, which may include chronological age, mortality-related phenotypic aging, pace of aging, or organ- and system-specific aging. They are therefore better viewed as complementary perspectives than interchangeable estimates of a single biological age (3).

DNA-methylation clocks are among the most studied. The Horvath clock estimates chronological age, DNAm PhenoAge reflects a mortality- and morbidity-related phenotypic-age measure, and DunedinPACE estimates the pace of biological aging rather than age itself (2,4,23). Clinical biomarker composites and proteomic clocks use blood-based measures to estimate systemic or organ-specific aging. These approaches have substantial research validation but require biological sample collection and laboratory processing (3).

Imaging- and sensor-based approaches offer another way to estimate aging, although their methods, validation, and interpretation vary. Importantly, the imaging modality alone does not determine what type of aging is being measured. Some imaging clocks estimate chronological age, with the difference between predicted and actual age reported as an age gap, while others are trained against a health-related phenotype or system-specific risk construct. Retinal models illustrate this distinction: retinal age-gap models estimate chronological age, RetiPhenoAge estimates Phenotypic Age (24), and BioAge is based on a cardiovascular-risk construct. Their outputs are therefore not interchangeable simply because each is derived from retinal images and expressed in years.

Within this landscape, BioAge is a cardiovascular-focused, imaging-based biological-age measure rather than an estimate of whole-body biological age. No head-to-head evidence has established superiority over other biological-age measures, and results from models designed for different purposes should not be expected to agree. Table 1 summarizes the principal approaches and practical considerations.

Table 1. Overview of Common Biological Age Approaches

CKM, cardiovascular-kidney-metabolic; CT, computed tomography; DNAm, DNA methylation; ECG, electrocardiogram; MRI, magnetic resonance imaging.

5. Evidence Summary and Current Limitations

Evidence relevant to BioAge comes from three distinct streams: direct BioAge studies, evidence supporting the underlying cardiovascular-risk model, and independent research on other retinal-aging models. The streams are complementary but not interchangeable. Findings from another model support the broader premise of retinal aging analysis but do not directly validate BioAge.

5.1 Direct evidence for BioAge

Associations with cardiovascular-kidney-metabolic indicators. A 2026 study evaluated retinal BioAge in held-out UK Biobank and the US EyePACS cohorts comprising 5,476 and 9,786 participants, respectively, aged 41–70 years. In both populations, the highest quartile of BioAge gap generally had less favorable CKM biomarker profiles and a higher prevalence of several clinical CKM indicators than the lowest quartile (18). In UK Biobank, the highest BioAge gap quartile had higher systolic and diastolic blood pressure, arterial stiffness, HbA1c, body mass index, and waist circumference. Hypertension prevalence was 51.3% in the highest BioAge gap quartile compared with 30.0% in the lowest quartile. The prevalence of chronic kidney disease, diabetes, and obesity was also higher in the highest quartile. In EyePACS, higher BioAge gaps were associated with higher systolic blood pressure and HbA1c, and lower estimated glomerular filtration rate. Chronic kidney disease prevalence was 35.8% in the highest BioAge gap quartile compared with 20.8% in the lowest quartile. The prevalence of hypertension and suboptimally controlled diabetes was also higher in the highest quartile.

These findings directly associate BioAge gap with CKM burden across two demographically and clinically different populations. However, the analysis was cross-sectional and relied largely on highest-versus-lowest quartile comparisons. It does not establish prospective event prediction, added value beyond conventional assessment, or causality.

Associations with telomere length and cardiovascular biomarkers. A 2024 UK Biobank study found modest associations between higher BioAge and shorter leukocyte telomere length, as a marker of cellular aging, as well as cardiovascular biomarkers (19). This supports biological plausibility but does not establish clinical validity or predictive performance.

Associations with cardiovascular health. An additional analysis linked BioAge with American Heart Association’s Life’s Essential 8 framework (20). Participants with more favorable cardiovascular health scores had a lower mean BioAge gap than those with poorer cardiovascular health, with blood pressure accounting for much of the observed association.

5.2 Evidence supporting the cardiovascular construct

The underlying retinal deep-learning model was developed to estimate a Pooled Cohort Equations–derived 10-year ASCVD risk score and validated using UK Biobank and EyePACS 10K data (17). The results of a prospective pivotal clinical trial in a diverse US population were recently presented, showing the AI model achieved 91% sensitivity and 86% specificity for identifying individuals with an elevated ASCVD risk score (25,26). This work supports the cardiovascular basis of the BioAge model, but it does not show that BioAge itself predicts cardiovascular events or improves clinical outcomes.

In another study of more than 50,000 UK Biobank participants, retinal AI–predicted blood pressure had a modestly stronger association with future cardiovascular events than blood pressure measured at the same visit, although both associations were small (0.067 versus 0.049) (27). This supports the plausibility of cumulative retinal vascular signals but does not directly validate BioAge.

5.3 Independent evidence from retinal-aging models

Independent studies further support the extraction of aging-related information from retinal images. RetiPhenoAge, which estimates Phenotypic Age rather than chronological age or cardiovascular risk, was associated in UK Biobank with cardiovascular, cancer, and all-cause mortality, as well as with cardiovascular events, after adjustment for conventional risk factors (24). Participants in the highest quartile had adjusted hazard ratios of 1.92 for all-cause mortality and 1.70 for cardiovascular events versus the lowest quartile. The model was also evaluated in Singapore and US cohorts.

Other studies have expanded retinal-aging research by evaluating models across populations and developing retinal-aging clocks from longitudinal fundus imaging (21,28). A recent scoping review also summarized associations between retinal age measures and mortality and chronic disease (29). Collectively, they support the broader premise that retinal images contain aging-related health information. Because the models use different targets and algorithms, however, they do not directly validate BioAge.

5.4 Summary of Current Evidence and Remaining Gaps

Together, the available studies support the cardiovascular and biological relevance of BioAge. Table 2 summarizes the principal findings and the claims they support.

Table 2. Principal BioAge Findings and Their Interpretation

ASCVD, atherosclerotic cardiovascular disease; CKM, cardiovascular-kidney-metabolic; PCE, Pooled Cohort Equations.

Current evidence associates BioAge with cardiovascular, kidney, metabolic, and aging-related health measures across populations with different demographic and clinical characteristics. Although several analyses used UK Biobank data, the recent CKM study also included a large, predominantly Hispanic US population with diabetes. However, the studies are largely cross-sectional, and have not established prospective event prediction, longitudinal responsiveness, or improved outcomes. Additional studies across broader age ranges, populations, imaging devices, and real-world settings are needed to define generalizability and clinical application. These gaps are consistent with broader challenges with biological clocks, including standardization and individual-level interpretation (3).

6. Interpreting BioAge Results

To support clear and consistent conversations, BioAge gap results can be described using three communication ranges as shown in Table 3. Suggested discussion is focused around the American Heart Association’s Life’s Essential 8, which provides a practical structure for preventive-health discussions: healthy eating, physical activity, avoidance of tobacco and nicotine, healthy sleep, and management of weight, blood cholesterol, blood glucose, and blood pressure (30).

Table 3. Suggested Communication Framework for BioAge Gap Results

The proposed ranges are intended as communication aids rather than validated clinical categories or treatment thresholds. They should be interpreted alongside established clinical information and should not independently determine clinical management.

References

1. Jylhävä J, Pedersen NL, Hägg S. Biological Age Predictors. EBioMedicine. 2017;21:29–36.
2. Levine ME, Lu AT, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY). 2018;10(4):573–91.
3. Wyss-Coray T, Topol EJ. Biological aging clocks in health and disease. Nat Med. 2026;32(7):2383–94.
4. Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. Elife. 2022;11:e73420.
5. Flammer J, Konieczka K, Bruno RM, Virdis A, Flammer AJ, Taddei S. The eye and the heart. Eur Heart J. 2013;34(17):1270–8.
6. London A, Benhar I, Schwartz M. The retina as a window to the brain-from eye research to CNS disorders. Nat Rev Neurol. 2013;9(1):44–53.
7. Hanssen H, Streese L, Vilser W. Retinal vessel diameters and function in cardiovascular risk and disease. Prog Retin Eye Res. 2022;91:101095.
8. Seidelmann SB, Claggett B, Bravo PE, et al. Retinal Vessel Calibers in Predicting Long-Term Cardiovascular Outcomes: The Atherosclerosis Risk in Communities Study. Circulation. 2016;134(18):1328–38.
9. Wong TY, Klein R, Couper DJ, et al. Retinal microvascular abnormalities and incident stroke: the Atherosclerosis Risk in Communities Study. Lancet. 2001;358(9288):1134–40.
10. McGeechan K, Liew G, Macaskill P, et al. Prediction of incident stroke events based on retinal vessel caliber: a systematic review and individual-participant meta-analysis. Am J Epidemiol. 2009;170(11):1323–32.
11. Wong TY, Mitchell P. Hypertensive retinopathy. N Engl J Med. 2004;351(22):2310–7.
12. Writing Committee Members*, Jones DW, Ferdinand KC, et al. 2025 AHA/ACC/AANP/AAPA/ABC/ACCP/ACPM/AGS/AMA/ASPC/NMA/PCNA/SGIM Guideline for the Prevention, Detection, Evaluation and Management of High Blood Pressure in Adults: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Hypertension. 2025;82(10):e212–316.
13. Wagner SK, Fu DJ, Faes L, et al. Insights into Systemic Disease through Retinal Imaging-Based Oculomics. Transl Vis Sci Technol. 2020;9(2):6.
14. Poplin R, Varadarajan AV, Blumer K, et al. Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nat Biomed Eng. 2018;2(3):158–64.
15. Rim TH, Lee G, Kim Y, et al. Prediction of systemic biomarkers from retinal photographs: development and validation of deep-learning algorithms. Lancet Digit Health. 2020;2(10):e526–36.
16. Hu W, Yii FSL, Chen R, et al. A Systematic Review and Meta-Analysis of Applying Deep Learning in the Prediction of the Risk of Cardiovascular Diseases From Retinal Images. Transl Vis Sci Technol. 2023;12(7):14.
17. Vaghefi E, Squirrell D, Yang S, et al. Development and validation of a deep-learning model to predict 10-year atherosclerotic cardiovascular disease risk from retinal images using the UK Biobank and EyePACS 10K datasets. Cardiovasc Digit Health J. 2024;5(2):59–69.
18. Squirrell D, Nielsen C, Vaghefi E, et al. Retinal BioAge is associated with indicators of cardiovascular-kidney-metabolic syndrome in UK and US populations. Sci Rep. 2026;16(1):10445.
19. Vaghefi E, An S, Corbett R, Squirrell D. Association of retinal image-based, deep learning cardiac BioAge with telomere length and cardiovascular biomarkers. Optom Vis Sci. 2024;101(7):464–9.
20. Squirrell D, Nielsen C, Vaghefi E, et al. Impact of Life’s Essential 8 Cardiovascular Health Score and Hypertension on Retinal Biological Aging. medRxiv. 2025:2025.10.16.25338198.
21. Yu Z, Chen R, Gui P, et al. A cross population study of retinal aging biomarkers with longitudinal pre-training and label distribution learning. npj Digit Med. 2025;8(1):344.
22. Wong TY, Hubbard LD, Klein R, et al. Retinal microvascular abnormalities and blood pressure in older people: the Cardiovascular Health Study. Br J Ophthalmol. 2002;86(9):1007–13.
23. Horvath S. DNA methylation age of human tissues and cell types. Genome Biol. 2013;14(10):R115.
24. Nusinovici S, Rim TH, Li H, et al. Application of a deep-learning marker for morbidity and mortality prediction derived from retinal photographs: a cohort development and validation study. Lancet Healthy Longev. 2024;5(10):100593.
25. Toku Eyes Ltd. CLAiR Pivotal Study. ClinicalTrials.gov. 2025. Accessed August 14, 2026. https://clinicaltrials.gov/study/NCT06808334
26. American College of Cardiology. CLAiR System: Can AI Reliably Assess CV Risk From Retinal Images? 2026. Accessed August 14, 2026. https://www.acc.org/latest-in-cardiology/articles/2026/03/25/21/27/mon-1045am-ai-analysis-acc-2026
27. Squirrell DM, Yang S, Xie L, et al. Blood Pressure Predicted From Artificial Intelligence Analysis of Retinal Images Correlates With Future Cardiovascular Events. JACC Adv. 2024;3(12):101410.
28. Ahadi S, Wilson KA, Babenko B, et al. Longitudinal fundus imaging and its genome-wide association analysis provide evidence for a human retinal aging clock. Elife. 2023;12:e82364.
29. Grimbly MJ, Koopowitz SM, Chen R, et al. Estimating biological age from retinal imaging: a scoping review. BMJ Open Ophthalmol. 2024;9(1):e001794.
30. Lloyd-Jones DM, Allen NB, Anderson CAM, et al. Life’s Essential 8: Updating and Enhancing the American Heart Association’s Construct of Cardiovascular Health: A Presidential Advisory From the American Heart Association. Circulation. 2022;146(5):e18–43.

© 2026 Topcon Healthcare, Inc. | MCA #6570