Accelerated biological aging, as well as cardiovascular, kidney, and metabolic (CKM) diseases, contribute to shortened healthspan. We studied a deep-learning model, retinal BioAge, and multiple indicators of CKM syndrome in participants from UK Biobank and the US-based EyePACS dataset. Retinal BioAge was trained on 77,887 retinal images and then used to analyze separate retinal images from UK Biobank (10,976) and EyePACS (19,856). In both datasets, CKM biomarker profiles were significantly worse for the top vs. bottom quartiles of BioAgeGap (retinal BioAge—chronological age), including measures of blood pressure, kidney function, adiposity, and glycemia. The top BioAgeGap quartile also had a significantly higher prevalence of clinical CKM indicators, including hypertension, kidney disease, and diabetes (UK Biobank) or suboptimally controlled diabetes (EyePACS). Thus, analysis of retinal images for accelerated biological aging may provide opportunistic screening to help identify individuals who could benefit from formal CKM assessment, potentially contributing to earlier detection and management of CKM syndrome.

We aimed to describe a 2-year outcome of eyes managed by practitioners benchmarked using a funnel plot by their frequency of treatment using vascular endothelial growth factor (VEGF) inhibitors for naive retinal vein occlusion (RVO). A multicentre, international, observational study of 29 doctors in 12 countries managing 1110 eyes with RVO commencing VEGF inhibitors between 1 January 2012–2022 tracked in the Fight Retinal Blindness! registry. We identified 3 outlying ‘intensive’ practitioners (managing 350/1110 eyes [32%]), 22 ‘typical’ practitioners (604/1110, [54%]) and 4 outlying ‘relaxed’ practitioners (156/1110, [14%]) with respective 24-month outcomes in Branch and Central RVO including the primary outcome, mean adjusted change in visual acuity (VA) in BRVO: +16.2, +13.6, +9.3 letters ( p < 0.01) and CRVO: +14.2, +12.7, +4.8 letters ( p < 0.01); adjusted change in macular thickness in BRVO −179, −150, −159 μm ( p < 0.01) and CRVO −324, −283, −232 μm ( p < 0.01); time-in-range with VA > 68 letters in BRVO 90, 78, 68 weeks ( p < 0.01) and CRVO 69, 60, 54 weeks ( p = 0.04); median injections 18, 13 and 10; median final injection intervals, BRVO 6, 9, 10 weeks and CRVO 6, 9 and 12 weeks; with no significant difference in adverse outcomes. At 24 months, the intensive practitioners were treating RVO using VEGF inhibitors with twice the frequency of the relaxed practitioners; however, their patients had gained twice (BRVO) to three times (CRVO) more letters of VA.

Patients continue to self-present to ophthalmology with advanced diabetic retinopathy. An audit of people living with diabetes attending our regional diabetes clinic revealed a significant number had undetected vision-threatening diabetic retinopathy despite regular community optometry review. Further work is required to determine why and whether this is a more widespread issue.

Background: Abnormalities in the retina have a profound impact on vision, and accurate diagnosis and monitoring are essential for effective clinical management. Retinal hyperreflective foci (HRF), lesions, or dots, identified using optical coherence tomography (OCT), are observed in both animals and humans and have been associated with several ocular conditions, including diabetic retinopathy (DR), age-related macular degeneration (AMD), and retinal vascular diseases. Methods: To evaluate the relevance of retinal HRF, we conducted a comprehensive scoping review of the literature published up to July 2024 including in the discussion key papers that emerged in 2025. Our search spanned electronic databases utilizing carefully identified search terms related to HRF and OCT within the last six years. We excluded publications on HRF outside the retina, treatments, non-peer-reviewed content, duplicates, studies older than 6 years, and those not focused on AMD, DR, or glaucoma. Results: A total of 141,085 records were initially identified from various databases and further refined based on keywords and content relevance. Finally, 42 reports meeting the criteria were retained for in-depth analysis. HRF were observed mainly in OCT scans of the AMD retina, as well as in DR and, to a lesser extent, in other retinopathies and interestingly in glaucoma. In AMD, HRF are described as a marker for disease progression, often associated with a compromised photoreceptor structure. In DR, HRF indicated issues such as abnormal blood vessels and cellular changes linked to microglia activation. In glaucoma, HRF may reflect microglia and macrophage activation. Most publications concur that the presence of HRF correlates with inflammatory processes and aging in the retina, with early appearance of small HRF serving as a biomarker for ocular disease. The size of HRF and their location were consistent with disease presentation. Conclusion: There is an agreement that HRF of less than 30 μm are biomarkers of inflammation in the retina despite having variable intraretinal locations. HRF resulting from the effect of aging can be discerned from AMD based on their quantity and appearance. The results show the importance of HRF as a biomarker of ocular disease and confirm that HRF are indicative of an inflammatory eye disorder.

Recent developments in artificial intelligence (AI) have seen a proliferation of algorithms that are now capable of predicting a range of systemic diseases from retinal images. Unlike traditional retinal disease detection AI models which are trained on well-recognised retinal biomarkers, systemic disease detection or "oculomics" models use a range of often poorly characterised retinal biomarkers to arrive at their predictions. As the retinal phenotype that oculomics models use may not be intuitive, clinicians have to rely on the developers' explanations of how these algorithms work in order to understand them. The discipline of understanding how AI algorithms work employs two similar but distinct terms: Explainable AI and Interpretable AI (iAI). Explainable AI describes the holistic functioning of an AI system, including its impact and potential biases. Interpretable AI concentrates solely on examining and understanding the workings of the AI algorithm itself. iAI tools are therefore what the clinician must rely on if they are to understand how the algorithm works and whether its predictions are reliable. The iAI tools that developers use can be delineated into two broad categories: Intrinsic methods that improve transparency through architectural changes and post-hoc methods that explain trained models via external algorithms. Currently post-hoc methods, class activation maps in particular, are far more widely used than other techniques but they have their limitations especially when applied to oculomics AI models. Aimed at clinicians, we examine how the key iAI methods work, what they are designed to do and what their limitations are when applied to oculomics AI. We conclude by discussing how combining existing iAI techniques with novel approaches could allow AI developers to better explain how their oculomics models work and reassure clinicians that the results issued are reliable.

Withdrawal Statement“The authors have withdrawn their manuscript owing to major updates that will be made prior to resubmission. Therefore, the authors do not wish this work to be cited as reference for the project. If you have any questions, please contact the corresponding author.”

BACKGROUND Atherosclerotic cardiovascular disease (ASCVD) is a leading cause of death globally, and early detection of high -risk individuals is essential for initiating timely interventions. The authors aimed to develop and validate a deep learning (DL) model to predict an individual 's elevated 10-year ASCVD risk score based on retinal images and limited demographic data. METHODS The study used 89,894 retinal fundus images from 44,176 UK Biobank participants (96% non-Hispanic White, 5% diabetic) to train and test the DL model. The DL model was developed using retinal images plus age, race/ethnicity, and sex at birth to predict an individual 's 10-year ASCVD risk score using the pooled cohort equation (PCE) as the ground truth. This model was then tested on the US EyePACS 10K dataset (5.8% non-Hispanic White, 99.9% diabetic), composed of 18,900 images from 8969 diabetic individuals. Elevated ASCVD risk was de fi ned as a PCE score of > 7.5%. RESULTS In the UK Biobank internal validation dataset, the DL model achieved an area under the receiver operating characteristic curve of 0.89, sensitivity 84%, and speci fi city 90%, for detecting individuals with elevated ASCVD risk scores. In the EyePACS 10K and with the addition of a regression-derived diabetes modi fi er, it achieved sensitivity 94%, speci fi city 72%, mean error -0.2%, and mean absolute error 3.1%. CONCLUSION This study demonstrates that DL models using retinal images can provide an additional approach to estimating ASCVD risk, as well as the value of applying DL models to different external datasets and opportunities about ASCVD risk assessment in patients living with diabetes.

Deep learning and artificial neural networks have been extensively applied to the automated diagnosis of retinal diseases from fundus images. Recent advancements have also led researchers to leverage deep learning to examine the connections between the retina and systemic health in a discipline termed oculomics. However, as oculomics models likely combine multiple retinal features to arrive at their conclusions, traditional methods in model interpretation, such as attribution saliency maps, often provide uncompelling and open-ended explanations that are prone to interpretation bias, highlighting a need for the examination of alternative strategies that can quantitatively describe model behavior. One potential solution is neuron activation patterns, which were previously applied to real-time fault diagnosis of deep learning models. In this study, we proposed a novel and experimental framework of neuron activation pattern synthesis leveraging image similarity metrics, with the outcome being a continuous, metric-based descriptor of underlying model behavior. We applied our approach in examining a model predicting systolic blood pressure from fundus images trained on the United Kingdom Biobank dataset. Our results show that the metric-based descriptor was meaningfully related to cardiovascular risk, a real-life outcome that can be expected to be related to blood pressure-related biomarkers identified from a fundus image. Furthermore, it was also able to uncover two biologically distinct and statistically significant groups among participants who were assigned the same predicted outcome and whose distinctness would otherwise be imperceivable without the insights generated by our approach. These results demonstrate the feasibility of this prototypical approach in neuron activation pattern synthesis for oculomics models. Further work is now required to validate these results on external datasets.

PURPOSE:To evaluate the 3-year outcomes of VEGF inhibitors in the treatment of cystoid macular edema due to branch retinal vein occlusion (BRVO) in an international multicenter cohort of eyes. DESIGN:Multicenter, international, BRVO database study. SUBJECTS:Seven hundred forty-seven patients (760 eyes) undergoing intravitreal therapy for BRVO for 3 years in a multicenter international setting. METHODS:Demographics, visual acuity (VA) in logarithm of the minimum angle of resolution letters, central subfield thickness (CST), treatments, number of injections, and visits data was collected using a validated web-based tool. MAIN OUTCOME MEASURES:Visual acuity gain at 3 years in logarithm of the minimum angle of resolution letters. Secondary outcome measures included anatomical results, treatment pattern, and percentage of completers. A subgroup analysis by study drug was conducted for clinical outcomes. RESULTS:Mean adjusted VA change was +11 letters (95% confidence interval 9-13), mean adjusted change in CST was -176 μm (-193, -159). Median number of injections/visits was 16 of 24 at 3 years of follow-up. Most eyes received VEGF inhibitors exclusively (89%, n = 677) and as a monotherapy in 71% (n = 538). Few eyes were switched to steroids (11%, n = 83). Suspensions in treatment >180 days occurred in 26% of study eyes. Aflibercept showed greater CST reductions (-147 vs. -128 vs. -114 μm; P < 0.001) and significantly lower switching rates (14% vs. 38% vs. 33%; P < 0.001) compared with ranibizumab and bevacizumab, respectively. CONCLUSIONS:This international study of 3-year BRVO outcomes after starting treatment with VEGF inhibitors found adequate visual and anatomical results in routine clinical care. Visual outcomes were similar among the different initiating VEGF inhibitors, although eyes starting with aflibercept had better anatomical outcomes and a lower switching rate. FINANCIAL DISCLOSURE(S):Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

SIGNIFICANCE:Our retinal image-based deep learning (DL) cardiac biological age (BioAge) model could facilitate fast, accurate, noninvasive screening for cardiovascular disease (CVD) in novel community settings and thus improve outcome with those with limited access to health care services. PURPOSE:This study aimed to determine whether the results issued by our DL cardiac BioAge model are consistent with the known trends of CVD risk and the biomarker leukocyte telomere length (LTL), in a cohort of individuals from the UK Biobank. METHODS:A cross-sectional cohort study was conducted using those individuals in the UK Biobank who had LTL data. These individuals were divided by sex, ranked by LTL, and then grouped into deciles. The retinal images were then presented to the DL model, and individual's cardiac BioAge was determined. Individuals within each LTL decile were then ranked by cardiac BioAge, and the mean of the CVD risk biomarkers in the top and bottom quartiles was compared. The relationship between an individual's cardiac BioAge, the CVD biomarkers, and LTL was determined using traditional correlation statistics. RESULTS:The DL cardiac BioAge model was able to accurately stratify individuals by the traditional CVD risk biomarkers, and for both males and females, those issued with a cardiac BioAge in the top quartile of their chronological peer group had a significantly higher mean systolic blood pressure, hemoglobin A 1c , and 10-year Pooled Cohort Equation CVD risk scores compared with those individuals in the bottom quartile (p<0.001). Cardiac BioAge was associated with LTL shortening for both males and females (males: -0.22, r2 = 0.04; females: -0.18, r2 = 0.03). CONCLUSIONS:In this cross-sectional cohort study, increasing CVD risk whether assessed by traditional biomarkers, CVD risk scoring, or our DL cardiac BioAge, CVD risk model, was inversely related to LTL. At a population level, our data support the growing body of evidence that suggests LTL shortening is a surrogate marker for increasing CVD risk and that this risk can be captured by our novel DL cardiac BioAge model.

Purpose To evaluate the 3-year outcomes of vascular endothelial growth factor (VEGF) inhibitors in the treatment of cystoid macular oedema (CME) due to branch retinal vein occlusion (BRVO) in an international multicenter cohort of eyes. Design Multicenter, international, BRVO database study. Subjects Seven hundred forty-seven patients (760 eyes) undergoing intravitreal therapy for BRVO for 3 years in a multicenter international setting. Methods Demographics, visual acuity (VA) in logarithm of the minimum angle of resolution (logMAR) letters, central subfield thickness (CST), treatments, number of injections and visits data was collected using a validated web-based tool. Main outcome measures Visual acuity (VA) gain at 3 years in LogMAR letters. Secondary outcome measures included anatomical results, treatment pattern and percentage of completers. A subgroup analysis by study drug was conducted for clinical outcomes. Results Mean adjusted VA change was +11 letters (95% CI 9,13), mean adjusted change in CST was -176μm (-193, -159). Median number of injections/visits was 16/24 at 3 years of follow-up. Most eyes received VEGF inhibitors exclusively (89%, n=677) and as a monotherapy in 71% (n=538). Few eyes were switched to steroids (11%, n=83). Suspensions in treatment >180 days occurred in 26% of study eyes. Aflibercept showed greater CST reductions (-147 vs -128 vs -114μm; p< 0.001) and significantly lower switching rates (14% vs 38% vs 33%, p< 0.001) compared with ranibizumab and bevacizumab, respectively. Conclusions This international study of 3-year BRVO outcomes after starting treatment with VEGF inhibitors found adequate visual and anatomical results in routine clinical care. Visual outcomes were similar amongst the different initiating VEGF inhibitors, although eyes starting with aflibercept had better anatomical outcomes and a lower switching rate.

To evaluate the 3-year outcomes of VEGF inhibitors in the treatment of cystoid macular edema due to branch retinal vein occlusion (BRVO) in an international multicenter cohort of eyes.

Background: We aimed to describe a 2-year outcome of eyes managed by practitioners benchmarked using a funnel plot by their frequency of treatment using vascular endothelial growth factor (VEGF) inhibitors for naive retinal vein occlusion (RVO). Methods: A multicentre, international, observational study of 29 doctors in 12 countries managing 1110 eyes with RVO commencing VEGF inhibitors between 1 January 2012-2022 tracked in the Fight Retinal Blindness! registry. Results: We identified 3 outlying 'intensive' practitioners (managing 350/1110 eyes [32%]), 22 'typical' practitioners (604/1110, [54%]) and 4 outlying 'relaxed' practitioners (156/1110, [14%]) with respective 24-month outcomes in Branch and Central RVO including the primary outcome, mean adjusted change in visual acuity (VA) in BRVO: +16.2, +13.6, +9.3 letters (p < 0.01) and CRVO: +14.2, +12.7, +4.8 letters (p < 0.01); adjusted change in macular thickness in BRVO -179, -150, -159 mu m (p < 0.01) and CRVO -324, -283, -232 mu m (p < 0.01); time-in-range with VA > 68 letters in BRVO 90, 78, 68 weeks (p < 0.01) and CRVO 69, 60, 54 weeks (p = 0.04); median injections 18, 13 and 10; median final injection intervals, BRVO 6, 9, 10 weeks and CRVO 6, 9 and 12 weeks; with no significant difference in adverse outcomes. Conclusions: At 24 months, the intensive practitioners were treating RVO using VEGF inhibitors with twice the frequency of the relaxed practitioners; however, their patients had gained twice (BRVO) to three times (CRVO) more letters of VA.

PURPOSE:To compare 1-year outcomes of eyes with diabetic macular edema (DME) treated in routine clinical practice based on the proportion of visits where intravitreal VEGF inhibitor injections were delivered. DESIGN:Cohort study. PARTICIPANTS:There were 2288 treatment-naive eyes with DME starting intravitreal VEGF inhibitor therapy from October 31, 2015 to October 31, 2021 from the Fight Retinal Blindness! international outcomes registry. METHODS:Eyes were grouped according to the proportion of visits at which an injection was received, Group A with less than the median of 67% (n = 1172) versus Group B with greater than the median (n = 1116). MAIN OUTCOME MEASURES:Mean visual acuity (VA) change after 12 months of treatment. RESULTS:The mean (95% confidence interval [CI]) VA change after 12 months of treatment was 3.6 (2.8-4.4) letters for eyes in Group A versus 5.2 (4.4-5.9) letters for eyes in Group B (P = 0.005). The mean (95% CI) central subfield thickness (CST) change was -69 (-76 to -61) μm and -85 (-92 to -78) μm for eyes in Group A versus Group B, respectively (P = 0.002). A moderate positive correlation was observed between the number of injections received over 12 months of treatment and the change in VA (P < 0.001). Additionally, eyes that received more injections had a moderately greater CST reduction. CONCLUSIONS:This registry analysis found that overall VA and anatomic outcomes tended to be better in DME eyes treated at a greater proportion of visits in the first year of intravitreal VEGF inhibitor therapy. FINANCIAL DISCLOSURE(S):Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Background: There is a growing recognition of the divergence between biological and chronological age, as well as the interaction among cardiovascular, kidney, and metabolic (CKM) diseases, known as CKM syndrome, in shortening both lifespan and healthspan. Detecting indicators of CKM syndrome can prompt lifestyle and risk-factor management to prevent progression to adverse clinical events. In this study, we tested a novel deep-learning model, retinal BioAge, to determine whether it could identify individuals with a higher prevalence of CKM indicators compared to their peers of similar chronological age. Methods: Retinal images and health records were analyzed from both the UK Biobank population health study and the US-based EyePACS 10K dataset of persons living with diabetes. 77,887 retinal images from 44,731 unique participants were used to train the retinal BioAge model. For validation, separate test sets of 10,976 images (5,476 individuals) from UK Biobank and 19,856 retinal images (9,786 individuals) from EyePACS 10K were analyzed. Retinal AgeGap (retinal BioAge — chronological age) was calculated for each participant, and those in the top and bottom retinal AgeGap quartiles were compared for prevalence of abnormal blood pressure, cholesterol, kidney function, and hemoglobin A1c. Results: In UK Biobank, participants in the top retinal AgeGap quartile had significantly higher prevalence of hypertension compared to the bottom quartile (36.3% vs. 29.0%, p<0.001), while the prevalence was similar for elevated non-HDL cholesterol (77.9% vs. 78.4%, p=0.80), impaired kidney function (4.8% vs. 4.2%, p=0.60), and diabetes (3.1% vs. 2.2%, p=0.24). In contrast, EyePACS 10K individuals in the top retinal AgeGap quartile had higher prevalence of elevated non-HDL cholesterol (49.9% vs. 43.0%, p<0.001), impaired kidney function (36.7% vs. 23.1%, p<0.001), suboptimally controlled diabetes (76.5% vs. 60.0%, p<0.001), and diabetic retinopathy (52.9% vs. 8.0%, p<0.001), but not hypertension (53.8% vs. 55.4%, p=0.33). Conclusion: A deep-learning retinal BioAge model identified individuals who had a higher prevalence of underlying indicators of CKM syndrome compared to their peers, particularly in a diverse US dataset of persons living with diabetes. ### Competing Interest Statement EV, SA, SM, SY, LX, DS, and MVM report employment by Toku Eyes. MVM reports compensation by Porter Health for consultant services. MKD and HH report employment by Topcon Healthcare. RNW reports compensation by Toku Eyes for consultant and board of directors services and by Topcon Healthcare for consultant services, as well as research instruments from Topcon, Visionix, Centervue, and Konan. ### Clinical Trial This was not a prospective clinical trial. ### Funding Statement Toku Eyes ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: UK Biobank (IRB UOA-86299) EyePACS 10K (IRB UCB 2017-09-10340) I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The datasets from UK Biobank and EyePACS are available through those organizations.

Recent advancements in artificial intelligence (AI) have prompted researchers to expand into the field of oculomics; the association between the retina and systemic health. Unlike conventional AI models trained on well-recognized retinal features, the retinal phenotypes that most oculomics models use are more subtle. Consequently, applying conventional tools, such as saliency maps, to understand how oculomics models arrive at their inference is problematic and open to bias. We hypothesized that neuron activation patterns (NAPs) could be an alternative way to interpret oculomics models, but currently, most existing implementations focus on failure diagnosis. In this study, we designed a novel NAP framework to interpret an oculomics model. We then applied our framework to an AI model predicting systolic blood pressure from fundus images in the United Kingdom Biobank dataset. We found that the NAP generated from our framework was correlated to the clinically relevant endpoint of cardiovascular risk. Our NAP was also able to discern two biologically distinct groups among participants who were assigned the same predicted systolic blood pressure. These results demonstrate the feasibility of our proposed NAP framework for gaining deeper insights into the functioning of oculomics models. Further work is required to validate these results on external datasets.

The aim of this study is to correlate small dot hyper‐reflective foci (HRF) observed in spectral domain optical coherence tomography (SD‐OCT) scans of an animal model of hyperglycaemia with focal electroretinography (fERG) response and immunolabelling of retinal markers. The eyes of an animal model of hyperglycaemia showing signs of diabetic retinopathy (DR) were imaged using SD‐OCT. Areas showing dot HRF were further evaluated using fERG. Retinal areas enclosing the HRF were dissected and serially sectioned, stained and labelled for glial fibrillary acidic protein (GFAP) and a microglial marker (Iba‐1). Small dot HRF were frequently seen in OCT scans in all retinal quadrants in the inner nuclear layer or outer nuclear layer in the DR rat model. Retinal function in the HRF and adjacent areas was reduced compared with normal control rats. Microglial activation was detected by Iba‐1 labelling and retinal stress identified by GFAP expression in Müller cells observed in discrete areas around small dot HRF. Small dot HRF seen in OCT images of the retina are associated with a local microglial response. This study provides the first evidence of dot HRF correlating with microglial activation, which may allow clinicians to better evaluate the microglia‐mediated inflammatory component of progressive diseases showing HRF.

Deep learning (DL) models have shown promise in detecting chronic kidney disease (CKD) from fundus photographs. However, previous studies have utilized a serum creatinine-only estimated glomerular rate (eGFR) equation to measure kidney function despite the development of more up-to-date methods. In this study, we developed two sets of DL models using fundus images from the UK Biobank to ascertain the effects of using a creatinine and cystatin-C eGFR equation over the baseline creatinine-only eGFR equation on fundus image-based DL CKD predictors. Our results show that a creatinine and cystatin-C eGFR significantly improved classification performance over the baseline creatinine-only eGFR when the models were evaluated conventionally. However, these differences were no longer significant when the models were assessed on clinical labels based on ICD10. Furthermore, we also observed variations in model performance and systemic condition incidence between our study and the ones conducted previously. We hypothesize that limitations in existing eGFR equations and the paucity of retinal features uniquely indicative of CKD may contribute to these inconsistencies. These findings emphasize the need for developing more transparent models to facilitate a better understanding of the mechanisms underpinning the ability of DL models to detect CKD from fundus images.

Purpose: To create an ensemble of Convolutional Neural Networks (CNNs), capable of detecting and stratifying the risk of progressive age-related macular degeneration (AMD) from retinal photographs. Methods: Three individual CNNs are trained to accurately detect 1) advanced AMD, 2) drusen size and 3) the presence or otherwise of pigmentary abnormalities, from macular centered retinal images were developed. The CNNs were then arranged in a "cascading" architecture to calculate the Age-related Eye Disease Study (AREDS) Simplified 5-level risk Severity score (Risk Score 0 - Risk Score 4), for test images. The process was repeated creating a simplified binary "low risk" (Scores 0-2) and "high risk" (Risk Score 3- Participants: There were a total of 188,006 images, of which 118,254 images were deemed gradable, representing 4591 patients, from the AREDS1 dataset. The gradable images were split into 50%/25%/25% ratios for training, validation and test purposes. Main Outcome Measures: The ability of the ensemble of CNNs using retinal images to predict an individual's risk of experiencing progression of their AMD based on the AREDS 5-step Simplified Severity Scale. Results: When assessed against the 5-step Simplified Severity Scale, the results generated by the ensemble of CNN's achieved an accuracy of 80.43% (quadratic kappa 0.870). When assessed against a simplified binary (Low Risk/High Risk) classification, an accuracy of 98.08%, sensitivity of >= 85% and specificity of >= 99% was achieved. Conclusion: We have created an ensemble of neural networks, trained on the AREDS 1 dataset, that is able to accurately calculate an individual's score on the AREDS 5-step Simplified Severity Scale for AMD. If the results presented were replicated, then this ensemble of CNNs could be used as a screening tool that has the potential to significantly improve health outcomes by identifying asymptomatic individuals who would benefit from AREDS2 macular supplements.