Search

Ashleigh Barrett-Young,1 Wickliffe C Abraham,1 Carol Y Cheung,2 Jesse Gale,3 Sean Hogan,1 David Ireland,1 Ross Keenan,4– 6 Annchen R Knodt,7 Tracy R Melzer,6,8,9 Terrie E Moffitt,7,10 Sandhya Ramrakha,1 Yih Chung Tham,11,12 Graham A Wilson,13 Tien Yin Wong,12,14 Ahmad R Hariri,7 Richie Poulton1 1Department of Psychology, University of Otago, Dunedin, New Zealand; 2Department of Ophthalmology and Visual Sciences, Faculty of Medicine, the Chinese University of Hong Kong, Hong Kong; 3Department of Surgery & Anaesthesia, University of Otago, Wellington, New Zealand; 4Department of Radiology, Christchurch Hospital, Christchurch, New Zealand; 5Pacific Radiology Group, Christchurch, New Zealand; 6New Zealand Brain Research Institute, Christchurch, New Zealand; 7Department of Psychology and Neuroscience, Duke University, Durham, NC, USA; 8Department of Medicine, University of Otago, Christchurch, New Zealand; 9School of Psychology, Speech and Hearing, University of Canterbury, Christchurch, New Zealand; 10Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK; 11Singapore Eye Research Institute, Singapore National Eye Centre, Duke-NUS Medical School, Singapore; 12Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore; 13Department of Medicine, University of Otago, Dunedin, New Zealand; 14Tsinghua Medicine, Tsinghua University, Beijing, People’s Republic of ChinaCorrespondence: Ashleigh Barrett-Young, Department of Psychology, University of Otago, PO Box 56, Dunedin, 9054, New Zealand, Email Ashleigh.barrett-young@otago.ac.nzPurpose: The retina has potential as a biomarker of brain health and Alzheimer’s disease (AD) because it is the only part of the central nervous system which can be easily imaged and has advantages over brain imaging technologies. Few studies have compared retinal and brain measurements in a middle-aged sample. The objective of our study was to investigate whether retinal neuronal measurements were associated with structural brain measurements in a middle-aged population-based cohort.Participants and Methods: Participants were members of the Dunedin Multidisciplinary Health and Development Study (n=1037; a longitudinal cohort followed from birth and at ages 3, 5, 7, 9, 11, 13, 15, 18, 21, 26, 32, 38, and most recently at age 45, when 94% of the living Study members participated). Retinal nerve fibre layer (RNFL) and ganglion cell-inner plexiform layer (GC-IPL) thickness were measured by optical coherence tomography (OCT). Brain age gap estimate (brainAGE), cortical surface area, cortical thickness, subcortical grey matter volumes, white matter hyperintensities, were measured by magnetic resonance imaging (MRI).Results: Participants with both MRI and OCT data were included in the analysis (RNFL n=828, female n=413 [49.9%], male n=415 [50.1%]; GC-IPL n=825, female n=413 [50.1%], male n=412 [49.9%]). Thinner retinal neuronal layers were associated with older brain age, smaller cortical surface area, thinner average cortex, smaller subcortical grey matter volumes, and increased volume of white matter hyperintensities.Conclusion: These findings provide evidence that the retinal neuronal layers reflect differences in midlife structural brain integrity consistent with increased risk for later AD, supporting the proposition that the retina may be an early biomarker of brain health.Keywords: optical coherence tomography, retina, Alzheimer’s, magnetic resonance imaging, brain aging

Background/aims: Large language models (LLMs) have substantial potential to enhance the efficiency of academic research. The accuracy and performance of LLMs in a systematic review, a core part of evidence building, has yet to be studied in detail., Methods: We introduced two LLM-based approaches of systematic review: an LLM-enabled fully automated approach (LLM-FA) utilising three different GPT-4 plugins (Consensus GPT, Scholar GPT and GPT web browsing modes) and an LLM-facilitated semi-automated approach (LLM-SA) using GPT4's Application Programming Interface (API). We benchmarked these approaches using three published systematic reviews that reported the prevalence of diabetic retinopathy across different populations (general population, pregnant women and children)., Results: The three published reviews consisted of 98 papers in total. Across these three reviews, in the LLM-FA approach, Consensus GPT correctly identified 32.7% (32 out of 98) of papers, while Scholar GPT and GPT4's web browsing modes only identified 19.4% (19 out of 98) and 6.1% (6 out of 98), respectively. On the other hand, the LLM-SA approach not only successfully included 82.7% (81 out of 98) of these papers but also correctly excluded 92.2% of 4497 irrelevant papers., Conclusions: Our findings suggest LLMs are not yet capable of autonomously identifying and selecting relevant papers in systematic reviews. However, they hold promise as an assistive tool to improve the efficiency of the paper selection process in systematic reviews., Competing Interests: Competing interests: TYW declares consulting fees from Aldropika Therapeutics, Bayer, Boehringer Ingelheim, Genentech, Iveric Bio, Novartis, Plano, Oxurion, Roche, Sanofi and Shanghai Henlius; funding from the National Key R&D Program, China (grant number 2022YFC2502802); and being an inventor, patent holder and co-founder of the start-up companies EyRiS and Visre. All other authors declare no competing interests., (© Author(s) (or their employer(s)) 2025. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ Group.)

Background: To evaluate the 6-year physiological rates-of-change in ganglion cell inner plexiform layer (GCIPL) and retinal nerve fibre layer (RNFL) thickness measured with optical coherence tomography., Methods: We included 2202 out of 2661 subjects from the population-based Singapore Chinese Eye Study who returned for follow-up 6 years after baseline examination (follow-up rate 87.7%). OCT scans with signal strength (SS) <6, imaging errors, and ocular pathologies were excluded. A linear mixed model was used to measure the rates-of-change in GCIPL and RNFL thickness. Time and difference between baseline and follow-up scan SS were modelled as fixed effect. Baseline age, baseline measurement, gender, hypertensive medication, diabetes status, cardiovascular disease, smoking status, body mass index, spherical equivalent (SE), intraocular pressure and optic disc area were each analysed in an interaction term with time., Results: The adjusted mean rate-of-change in average GCIPL was -0.312 μm/year in males and -0.235 μm/year in females. Older age and thicker GCIPL thickness at baseline were associated with higher rates-of-change while females and more hyperopic SE were associated with lower rates-of-change. The adjusted mean rate-of-change in average RNFL was -0.374 μm, with higher rates-of-change in the vertical quadrants and no differences between genders. Older age and thicker RNFL thickness at baseline were associated with higher rates-of-change in average RNFL and RNFL thickness in the vertical quadrants, and vice versa for each unit increase in scan SS and SE., Conclusion: Our population cohort provides data on physiological thinning of GCIPL and RNFL with age. Differentiating physiological changes in GCIPL and RNFL is important for more accurate clinical assessment., (© 2025 Royal Australian and New Zealand College of Ophthalmologists.)

Label noise is a common and important issue that would affect the model's performance in artificial intelligence. This study assessed the effectiveness and potential risks of automated label cleaning using an open-source framework, Cleanlab, in multi-category datasets of fundus photography and optical coherence tomography, with intentionally introduced label noise ranging from 0 to 70%. After six cycles of automatic cleaning, significant improvements are achieved in label accuracies (3.4-62.9%) and dataset quality scores (DQS, 5.1-74.4%). The majority (86.6 to 97.5%) of label errors were accurately modified, with minimal missed (0.5-2.8%) or misclassified (0.4-10.6%). The classification accuracy of RETFound significantly improved by 0.3-52.9% when trained with the datasets after cleaning. We also developed a DQS-guided cleaning strategy to mitigate over-cleaning. Furthermore, external validation on EyePACS and APTOS-2019 datasets boosted label accuracy by 1.3 and 1.8%, respectively. This approach automates label correction, enhances dataset reliability, and strengthens model performance efficiently and safely., Competing Interests: Competing interests: The authors declare no competing interests., (© 2025. The Author(s).)

Background: The American Heart Association recently published guidelines on how to clinically identify and categorize individuals with cardiovascular-kidney-metabolic (CKM) syndrome. The extent to which CKM syndrome prevalence and prognosis differ by sex remains unknown. This study aimed to examine the impact of sex on trends in prevalence over 30 years and the long-term prognosis of CKM syndrome in the United States., Methods: We analyzed nationally representative National Health and Nutrition Examination Survey 1988 to 2018 data collected from 33 868 US adults (aged ≥20 years) who were under surveillance for all-cause mortality through December 31, 2019. We examined the sex-specific prevalence of CKM syndrome and sex-specific CKM associations with all-cause mortality., Results: Of the 33 868 adults studied, the mean±SD age was 48.4±18.3 years with 52% women and 56% non-White. Overall prevalence of CKM syndrome increased steadily from 1988 to 2018 in both sexes, with a larger temporal rise in prevalent stage 3 CKM seen for men (from 18.9% to 22.4%) compared with women (from 13.9% to 15.2%). Over a median follow-up of 13.3 years, there were 8745 deaths. In the multivariable Cox regression analysis, worsening CKM severity was associated with all-cause mortality ( P <0.001 for both sexes), with greater magnitudes of risk seen in women (hazards ratio, 1.24-3.33) compared with men (hazards ratio, 0.85-2.60) across all stages (likelihood ratio test χ 2 , 19.0; P interaction <0.001); results were similar for cardiovascular mortality (likelihood ratio test χ 2 , 22.3; P interaction <0.001)., Conclusions: Women, compared with men, exhibited a lower prevalence of CKM stage 3 but experienced excess mortality risk across the spectrum of multisystem CKM dysfunction. These findings underscore the importance of identifying mechanisms underlying joint cardiovascular, kidney, and metabolic system pathophysiology to close a potentially widening sex disparities gap in multiorgan disease risk., Competing Interests: None.

Objectives: To determine the association between telomere length (TL) and age-related macular degeneration (AMD) and examine the potential variations with sex and ethnicity., Methods: Population-based, cross-sectional study. A total of 52,083 participants from the UK Biobank were included. Leucocyte TL, measured using quantitative polymerase chain reaction assay, was presented as the ratio of telomere repeat copy number relative to that of a single copy gene, and then log-transformed and Z-standardised. AMD cases were identified based on a combination of in-patient, self-reported and primary care data, and furtherly classified as early, intermediate and late AMD using the Beckmann classification system (based on more severe eye)., Results: Among the 52,083 participants aged 60.2 ± 5.4 years, 725 were any-AMD cases. AMD patients had shorter TL than those without AMD (-0.22 ± 0.95 vs. -0.10 ± 0.99, P = 0.001). In multivariable model, shorter TL (per standard deviation) was significantly associated with higher odds of AMD in Whites (OR:1.09; 95% CI: 1.01-1.18; P = 0.036). When stratified by sex and ethnicity, this association was only significant in White women (OR:1.14; 95%CI: 1.02, 1.27; P = 0.018), but not in men and nonwhite populations (all P ≥ 0.335). Among white women, the association was more pronounced (OR:1.47; 95%CI:1.23-1.77; P < 0.001) for intermediate/late AMD but not for early AMD (P = 0.789)., Conclusions: Shorter TL was associated with any AMD in white women but not in men and other ethnicities. Our findings highlight the potential role of telomere length in the pathogenesis of AMD and the importance of considering sex and ethnicity variation in this research area., Competing Interests: Competing interests: KT, CMGC and YCT are members of the Eye editorial board. The remaining authors declare no conflicts., (© 2024. The Author(s), under exclusive licence to The Royal College of Ophthalmologists.)

Background: Approximately 1 in 3 adults live with multiple chronic diseases. Digital health is being harnessed to improve continuity of care and management of chronic diseases. However, meaningful uptake of digital health for chronic disease management remains low. It is unclear how these innovations have been implemented and evaluated., Objective: This scoping review aims to identify how digital health innovations for chronic disease management have been implemented and evaluated: what implementation frameworks, methods, and strategies were used; how successful these strategies were; key barriers and enablers to implementation; and lessons learned and recommendations shared by study authors., Methods: We used the Joanna Briggs Institute methodology for scoping reviews. Five databases were searched for studies published between January 2015 and March 2023: PubMed, Scopus, CINAHL, PsycINFO, and IEEE Xplore. We included primary studies of any study design with any type of digital health innovations for chronic diseases that benefit patients, caregivers, or health care professionals. We extracted study characteristics; type of digital health innovation; implementation frameworks, strategies, and outcome measures used; barriers and enablers to implementation; lessons learned; and recommendations reported by study authors. We used established taxonomies to synthesize extracted data. Extracted barriers and enablers were grouped into categories for reporting. Descriptive statistics were used to consolidate extracted data., Results: A total of 252 studies were included, comprising mainly mobile health (107/252, 42.5%), eHealth (61/252, 24.2%), and telehealth (97/252, 38.5%), with some studies involving more than 1 innovation. Only 23 studies (23/252, 9.1%) reported using an implementation science theory, model, or framework; the most common were implementation theories, classic theories, and determinant frameworks, with 7 studies each. Of 252 studies, 144 (57.1%) used 2 to 5 implementation strategies. Frequently used strategies were "obtain and use patient or consumer feedback" (196/252, 77.8%); "audit and provide feedback" (106/252, 42.1%); and piloting before implementation or "stage implementation scale-up" (85/252, 33.7%). Commonly measured implementation outcomes were acceptability, feasibility, and adoption of the digital innovation. Of 252 studies, 247 studies (98%) did not measure service outcomes, while patient health outcomes were measured in 89 studies (35.3%). The main method used to assess outcomes was surveys (173/252, 68.7%), followed by interviews (95/252, 37.7%). Key barriers impacting implementation were data privacy concerns and patient preference for in-person consultations. Key enablers were training for health care workers and personalization of digital health features to patient needs., Conclusions: This review generated a summary of how digital health in chronic disease management is currently implemented and evaluated and serves as a useful resource for clinicians, researchers, health system managers, and policy makers planning real-world implementation. Future studies should investigate whether using implementation science frameworks, including how well they are used, would yield better outcomes compared to not using them., (©Candelyn Pong, Rachel Marjorie Wei Wen Tseng, Yih Chung Tham, Elaine Lum. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 12.12.2024.)

Purpose: Epidemiological studies and clinical trials have reported inconsistent findings regarding the protective role of omega-3 fatty acids in age-related macular degeneration (AMD), we investigated their association in a prospective cohort and examined causality using Mendelian randomization (MR) analyses., Design: Prospective cohort study and two-sample MR analyses., Participants: We included individuals of European descent from UK Biobank with plasma omega-3 and docosahexaenoic acid (DHA) measurement. Our MR analyses used genome-wide association study data on plasma level of omega-3 and DHA (UK Biobank, n=115,006) and AMD (dry, wet, and any) (FinnGen, n=208,690 to 209,122)., Methods: Cox regression models was used to assess the association between plasma omega-3/DHA levels and AMD incidence, adjusting for systemic covariates and AMD polygenetic risk score (PRS). Interaction effects of AMD genetic risk (PRS, CFH/ARMS2 genotypes) and plasma omega-3/DHA levels were tested. For MR analyses, we used random-effect inverse-variance weighted (RE-IVW) model as primary model and five other models for sensitivity. Causality between omega-3/DHA and AMD was considered significant if P-values were <0.05 in the primary model and at least 2 other models., Main Outcome Measures: The risk of AMD., Results: Among 258,350 AMD-free individuals, 5,068 (1.9%) developed AMD over 12.9-year follow-up. Higher plasma levels (in mmol/L) of omega-3 (HR, 0.80; 95%CI, 0.72, 0.95; P=0.006) and DHA (HR, 0.65; 95%CI, 0.44, 0.96; P=0.029) were associated with lower risk of being diagnosed with AMD. MR analyses showed causality between genetic predisposition to higher plasma omega-3 levels and lower risk of dry AMD (OR, 0.83; 95% CI, 0.73, 0.96, P=0.010), wet AMD (OR, 0.76; 95%CI, 0.65,0.88; P<0.001), and any AMD (OR, 0.82; 95%CI, 0.74, 0.92; P<0.001). Likewise, genetic predisposition to higher plasma DHA levels were causally associated with a lower risk of wet AMD (OR, 0.79; 95%CI, 0.65, 0.96; P= 0.017) and any AMD (OR, 0.84; 95%CI, 0.72, 0.98; P=0.030). No significant interaction was found between omega-3/DHA levels and AMD genetic risks (all P>0.05)., Conclusions: Our prospective cohort and MR analyses both suggested a protective effect of omega-3 and DHA on AMD, supporting the need for further clinical trials to test their effectiveness in AMD prevention and treatment., (Copyright © 2024. Published by Elsevier Inc.)

Purpose: Animal models suggest omega-3 polyunsaturated fatty acids (PUFAs) may protect against myopia by modulating choroidal blood perfusion, but clinical evidence is scarce and mixed. We aimed to determine the causality between omega-3 PUFAs and myopia using Mendelian randomization (MR) analysis., Design: Two-sample MR analysis., Methods: Exposures are genetically predicted plasma levels of 18 fatty acid (FA)-related traits. Spherical equivalent refraction (SER) and axial length were used as measurements of myopia. Genome-wide association study summary data on plasma levels of 18 FA-related traits (n=115,006), refractive spherical equivalent (n=351,091), axial length (n=69,945), and choroidal thickness (n=44,823) were sourced from the UK Biobank, the Genetic Epidemiology Research on Adult Health and Aging cohort, and the Consortium for Refractive Error and Myopia Study. We used 5 MR models and considered results statistically significant if the Bonferroni-corrected P value was ≤2.78 ×10 -3 in at least 3 MR models. The β represents the change in outcomes (SER in diopters; axial length in millimeters; and choroidal thickness in SD) per SD unit increase in FA levels., Results: At a Bonferroni-corrected significance, higher levels of omega-3 (β, 0.32-0.34), omega-3-total FA ratio (β, 0.31-0.44), docosahexaenoic acid (DHA) (β, 0.36-0.46), DHA-total FA ratio (β, 0.37-0.53), PUFA-total FA ratio (β, 0.07-1.003), and degree of unsaturation (β, 0.28-0.44) were associated with a more positive SER, suggesting a lower risk of myopia. Similar trends were observed for axial length albeit with borderline significance (P ≤ .035 in ≥2 models). Higher levels of omega-3, DHA, DHA-total FA ratio, PUFA-total FA ratio, PUFA-monounsaturated FA ratio, and degree of unsaturation were nominally associated with thicker choroidal thickness (β, 0.05-0.13; P ≤ .045 in ≥2 models)., Conclusion: Our multiple MR models suggest a protective effect of omega-3 and DHA on myopia, potentially through modulation of choroidal blood perfusion. Further randomized clinical trials are needed to confirm the effectiveness and determine the optimal dose and duration., (Copyright © 2024 Elsevier Inc. All rights reserved.)

We investigated whether the effect of lipid-lowering drugs (LLDs) on age-related macular degeneration (AMD) differs according to the main complement genetic variants in Singapore Epidemiology of Eye Diseases (SEED) ( n  = 5,579) and UK Biobank studies ( n  = 445,727). The effect of LLD was determined for each stratum of 20 complement genetic variants. In SEED, 484 individuals developed AMD and 216 showed progression over 6 years. In the UK Biobank, 913 participants developed AMD over 11 years. rs1061170 variant (complement factor H gene) was the only variant for which we found a protective effect in both populations. This effect was found in individuals carrying at least one C allele in SEED (odds ratio [OR] = 0.41; 95% confidence interval [CI], 0.19-0.87) and in individuals carrying two C alleles in UK Biobank (hazard ratio [HR] = 0.65; 95% CI, 0.45-0.93). These effects corresponded to a 50% and 35% decrease in AMD risk, respectively. Our study highlights the potential for personalized therapy for AMD based on complement genotyping., Competing Interests: The authors declare no competing interests., (© 2024 The Authors.)

Structural variants (SVs) are significant contributors to inter-individual genetic variation associated with traits and diseases. Current SV studies using whole-genome sequencing (WGS) have a largely Eurocentric composition, with little known about SV diversity in other ancestries, particularly from Asia. Here, we present a WGS catalogue of 73,035 SVs from 8392 Singaporeans of East Asian, Southeast Asian and South Asian ancestries, of which ~65% (47,770 SVs) are novel. We show that Asian populations can be stratified by their global SV patterns and identified 42,239 novel SVs that are specific to Asian populations. 52% of these novel SVs are restricted to one of the three major ancestry groups studied (Indian, Chinese or Malay). We uncovered SVs affecting major clinically actionable loci. Lastly, by identifying SVs in linkage disequilibrium with single-nucleotide variants, we demonstrate the utility of our SV catalogue in the fine-mapping of Asian GWAS variants and identification of potential causative variants. These results augment our knowledge of structural variation across human populations, thereby reducing current ancestry biases in global references of genetic variation afflicting equity, diversity and inclusion in genetic research., (© 2024. The Author(s).)

Importance: Myopic maculopathy (MM) is a major cause of vision impairment globally. Artificial intelligence (AI) and deep learning (DL) algorithms for detecting MM from fundus images could potentially improve diagnosis and assist screening in a variety of health care settings., Objectives: To evaluate DL algorithms for MM classification and segmentation and compare their performance with that of ophthalmologists., Design, Setting, and Participants: The Myopic Maculopathy Analysis Challenge (MMAC) was an international competition to develop automated solutions for 3 tasks: (1) MM classification, (2) segmentation of MM plus lesions, and (3) spherical equivalent (SE) prediction. Participants were provided 3 subdatasets containing 2306, 294, and 2003 fundus images, respectively, with which to build algorithms. A group of 5 ophthalmologists evaluated the same test sets for tasks 1 and 2 to ascertain performance. Results from model ensembles, which combined outcomes from multiple algorithms submitted by MMAC participants, were compared with each individual submitted algorithm. This study was conducted from March 1, 2023, to March 30, 2024, and data were analyzed from January 15, 2024, to March 30, 2024., Exposure: DL algorithms submitted as part of the MMAC competition or ophthalmologist interpretation., Main Outcomes and Measures: MM classification was evaluated by quadratic-weighted κ (QWK), F1 score, sensitivity, and specificity. MM plus lesions segmentation was evaluated by dice similarity coefficient (DSC), and SE prediction was evaluated by R2 and mean absolute error (MAE)., Results: The 3 tasks were completed by 7, 4, and 4 teams, respectively. MM classification algorithms achieved a QWK range of 0.866 to 0.901, an F1 score range of 0.675 to 0.781, a sensitivity range of 0.667 to 0.778, and a specificity range of 0.931 to 0.945. MM plus lesions segmentation algorithms achieved a DSC range of 0.664 to 0.687 for lacquer cracks (LC), 0.579 to 0.673 for choroidal neovascularization, and 0.768 to 0.841 for Fuchs spot (FS). SE prediction algorithms achieved an R2 range of 0.791 to 0.874 and an MAE range of 0.708 to 0.943. Model ensemble results achieved the best performance compared to each submitted algorithms, and the model ensemble outperformed ophthalmologists at MM classification in sensitivity (0.801; 95% CI, 0.764-0.840 vs 0.727; 95% CI, 0.684-0.768; P = .006) and specificity (0.946; 95% CI, 0.939-0.954 vs 0.933; 95% CI, 0.925-0.941; P = .009), LC segmentation (DSC, 0.698; 95% CI, 0.649-0.745 vs DSC, 0.570; 95% CI, 0.515-0.625; P < .001), and FS segmentation (DSC, 0.863; 95% CI, 0.831-0.888 vs DSC, 0.790; 95% CI, 0.742-0.830; P < .001)., Conclusions and Relevance: In this diagnostic study, 15 AI models for MM classification and segmentation on a public dataset made available for the MMAC competition were validated and evaluated, with some models achieving better diagnostic performance than ophthalmologists.

Alzheimer's disease (AD) is the leading cause of dementia worldwide. Current diagnostic modalities of AD generally focus on detecting the presence of amyloid β and tau protein in the brain (for example, positron emission tomography [PET] and cerebrospinal fluid testing), but these are limited by their high cost, invasiveness, and lack of expertise. Retinal imaging exhibits potential in AD screening and risk stratification, as the retina provides a platform for the optical visualization of the central nervous system in vivo, with vascular and neuronal changes that mirror brain pathology. Given the paradigm shift brought by advances in artificial intelligence and the emergence of disease-modifying therapies, this article aims to summarize and review the current literature to highlight 8 trends in an evolving landscape regarding the role and potential value of retinal imaging in AD screening., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

Aims: To develop and externally test deep learning (DL) models for assessing the image quality of three-dimensional (3D) macular scans from Cirrus and Spectralis optical coherence tomography devices., Methods: We retrospectively collected two data sets including 2277 Cirrus 3D scans and 1557 Spectralis 3D scans, respectively, for training (70%), fine-tuning (10%) and internal validation (20%) from electronic medical and research records at The Chinese University of Hong Kong Eye Centre and the Hong Kong Eye Hospital. Scans with various eye diseases (eg, diabetic macular oedema, age-related macular degeneration, polypoidal choroidal vasculopathy and pathological myopia), and scans of normal eyes from adults and children were included. Two graders labelled each 3D scan as gradable or ungradable, according to standardised criteria. We used a 3D version of the residual network (ResNet)-18 for Cirrus 3D scans and a multiple-instance learning pipline with ResNet-18 for Spectralis 3D scans. Two deep learning (DL) models were further tested via three unseen Cirrus data sets from Singapore and five unseen Spectralis data sets from India, Australia and Hong Kong, respectively., Results: In the internal validation, the models achieved the area under curves (AUCs) of 0.930 (0.885-0.976) and 0.906 (0.863-0.948) for assessing the Cirrus 3D scans and Spectralis 3D scans, respectively. In the external testing, the models showed robust performance with AUCs ranging from 0.832 (0.730-0.934) to 0.930 (0.906-0.953) and 0.891 (0.836-0.945) to 0.962 (0.918-1.000), respectively., Conclusions: Our models could be used for filtering out ungradable 3D scans and further incorporated with a disease-detection DL model, allowing a fully automated eye disease detection workflow., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)

Background: Biological ageing markers are useful to risk stratify morbidity and mortality more precisely than chronological age. In this study, we aimed to develop a novel deep-learning-based biological ageing marker (referred to as RetiPhenoAge hereafter) using retinal images and PhenoAge, a composite biomarker of phenotypic age., Methods: We used retinal photographs from the UK Biobank dataset to train a deep-learning algorithm to predict the composite score of PhenoAge. We used a deep convolutional neural network architecture with multiple layers to develop our deep-learning-based biological ageing marker, as RetiPhenoAge, with the aim of identifying patterns and features in the retina associated with variations of blood biomarkers related to renal, immune, liver functions, inflammation, and energy metabolism, and chronological age. We determined the performance of this biological ageing marker for the prediction of morbidity (cardiovascular disease and cancer events) and mortality (all-cause, cardiovascular disease, and cancer) in three independent cohorts (UK Biobank, the Singapore Epidemiology of Eye Diseases [SEED], and the Age-Related Eye Disease Study [AREDS] from the USA). We also compared the performance of RetiPhenoAge with two other known ageing biomarkers (hand grip strength and adjusted leukocyte telomere length) and one lifestyle factor (physical activity) for risk stratification of mortality and morbidity. We explored the underlying biology of RetiPhenoAge by assessing its associations with different systemic characteristics (eg, diabetes or hypertension) and blood metabolite levels. We also did a genome-wide association study to identify genetic variants associated with RetiPhenoAge, followed by expression quantitative trait loci mapping, a gene-based analysis, and a gene-set analysis. Cox proportional hazards models were used to estimate the hazard ratios (HRs) and corresponding 95% CIs for the associations between RetiPhenoAge and the different morbidity and mortality outcomes., Findings: Retinal photographs for 34 061 UK Biobank participants were used to train the model, and data for 9429 participants from the SEED cohort and for 3986 participants from the AREDS cohort were included in the study. RetiPhenoAge was associated with all-cause mortality (HR 1·92 [95% CI 1·42-2·61]), cardiovascular disease mortality (1·97 [1·02-3·82]), cancer mortality (2·07 [1·29-3·33]), and cardiovascular disease events (1·70 [1·17-2·47]), independent of PhenoAge and other possible confounders. Similar findings were found in the two independent cohorts (HR 1·67 [1·21-2·31] for cardiovascular disease mortality in SEED and 2·07 [1·10-3·92] in AREDS). RetiPhenoAge had stronger associations with mortality and morbidity than did hand grip strength, telomere length, and physical activity. We identified two genetic variants that were significantly associated with RetiPhenoAge (single nucleotide polymorphisms rs3791224 and rs8001273), and were linked to expression quantitative trait locis in various tissues, including the heart, kidneys, and the brain., Interpretation: Our new deep-learning-derived biological ageing marker is a robust predictor of mortality and morbidity outcomes and could be used as a novel non-invasive method to measure ageing., Funding: Singapore National Medical Research Council and Agency for Science, Technology and Research, Singapore., Competing Interests: Declaration of interests THR and GL own stocks in MediWhale, to whom RetiAge was licensed. ZZ holds a National Health and Medical Research Council Investigator Grant (2010072) and owns two patents for biological age prediction from ocular images (AU2023903213A0 and CN114782361A). T-YW is a consultant for Aldropika Therapeutics, Bayer, Boehringer Ingelheim, Carl Zeiss, Genentech, Iveric Bio, Novartis, Oxurion, Plano, Roche, Sanofi, and Shanghai Henlius; is an inventor, holds patents, and is a co-founder of the start-up companies EyRiS and Visre; and has interests in, and develops digital solutions for, eye diseases, including diabetic retinopathy, outside of the submitted work. C-YC holds licences and receives consultation fees from MediWhale. All other authors declare no competing interests., (Copyright © 2024 The Author(s). Published by Elsevier Ltd.. All rights reserved.)

Primary diabetes care and diabetic retinopathy (DR) screening persist as major public health challenges due to a shortage of trained primary care physicians (PCPs), particularly in low-resource settings. Here, to bridge the gaps, we developed an integrated image-language system (DeepDR-LLM), combining a large language model (LLM module) and image-based deep learning (DeepDR-Transformer), to provide individualized diabetes management recommendations to PCPs. In a retrospective evaluation, the LLM module demonstrated comparable performance to PCPs and endocrinology residents when tested in English and outperformed PCPs and had comparable performance to endocrinology residents in Chinese. For identifying referable DR, the average PCP's accuracy was 81.0% unassisted and 92.3% assisted by DeepDR-Transformer. Furthermore, we performed a single-center real-world prospective study, deploying DeepDR-LLM. We compared diabetes management adherence of patients under the unassisted PCP arm (n = 397) with those under the PCP+DeepDR-LLM arm (n = 372). Patients with newly diagnosed diabetes in the PCP+DeepDR-LLM arm showed better self-management behaviors throughout follow-up (P < 0.05). For patients with referral DR, those in the PCP+DeepDR-LLM arm were more likely to adhere to DR referrals (P < 0.01). Additionally, DeepDR-LLM deployment improved the quality and empathy level of management recommendations. Given its multifaceted performance, DeepDR-LLM holds promise as a digital solution for enhancing primary diabetes care and DR screening., (© 2024. The Author(s).)

Objective: Our objectives were to identify correlation patterns between complement and lipid pathways using a multiomics data integration approach and to determine how these interconnections affect age-related macular degeneration (AMD)., Design: Nested case-control study., Subjects and Controls: The analyses were performed in a subset of the Singapore Indian Eye Study. We randomly selected 155 AMD cases and age- and sex-matched them with 155 controls., Methods: Firstly, a multiomics data integration method was used to identify correlation patterns between the omics data. Then, we tested possible interactions between the lipids and complement proteins using logistic regression models., Main Outcome Measures: Age-related macular degeneration was determined according to the Beckman classification system. We measured in serum samples 35 complement proteins and 66 lipids, and used 9 genetic variants., Results: Among the 155 AMD cases, 93 (60.0%) had early and 62 (40.0%) intermediate AMD. Firstly, we identified 2 clusters between complement proteins and lipids involving (1) mannan-binding lectin serine protease 1 and several different high-density lipoprotein particles, and (2) complement factor H-related protein 1, carboxypeptidase N subunit 2 and complement component C8 gamma chain, and sphingomyelin and different cholesterol. Secondly, we identified 1 interaction between complement protein 1R and sphingomyelin with an odds of AMD 2 times higher for individuals with low levels of sphingomyelin and complement protein 1R (odds ratio = 2.13 [1.09, 4.17])., Conclusions: We report here, using a cutting-edge multiomics integration approach, the complex interconnections between genetic, metabolomics, and proteomic data. This method permitted us to obtain a holistic picture and identify multiomics signature of AMD pathophysiology. These results advocate for a personalized therapeutic approach that accounts for multiple pathways. However, these results need to be validated in larger studies with different ethnic groups., Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (© 2024 by the American Academy of Ophthalmology.)

Competing Interests: CCE, NLYC, MSF, JLWK, AM, TYC, NKY declare no competing interests.

Background/aims: To compare the performance of generative versus retrieval-based chatbots in answering patient inquiries regarding age-related macular degeneration (AMD) and diabetic retinopathy (DR)., Methods: We evaluated four chatbots: generative models (ChatGPT-4, ChatGPT-3.5 and Google Bard) and a retrieval-based model (OcularBERT) in a cross-sectional study. Their response accuracy to 45 questions (15 AMD, 15 DR and 15 others) was evaluated and compared. Three masked retinal specialists graded the responses using a three-point Likert scale: either 2 (good, error-free), 1 (borderline) or 0 (poor with significant inaccuracies). The scores were aggregated, ranging from 0 to 6. Based on majority consensus among the graders, the responses were also classified as 'Good', 'Borderline' or 'Poor' quality., Results: Overall, ChatGPT-4 and ChatGPT-3.5 outperformed the other chatbots, both achieving median scores (IQR) of 6 (1), compared with 4.5 (2) in Google Bard, and 2 (1) in OcularBERT (all p ≤8.4×10 -3 ). Based on the consensus approach, 83.3% of ChatGPT-4's responses and 86.7% of ChatGPT-3.5's were rated as 'Good', surpassing Google Bard (50%) and OcularBERT (10%) (all p ≤1.4×10 -2 ). ChatGPT-4 and ChatGPT-3.5 had no 'Poor' rated responses. Google Bard produced 6.7% Poor responses, and OcularBERT produced 20%. Across question types, ChatGPT-4 outperformed Google Bard only for AMD, and ChatGPT-3.5 outperformed Google Bard for DR and others., Conclusion: ChatGPT-4 and ChatGPT-3.5 demonstrated superior performance, followed by Google Bard and OcularBERT. Generative chatbots are potentially capable of answering domain-specific questions outside their original training. Further validation studies are still required prior to real-world implementation., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)

Background: Large language models (LLMs) are fast emerging as potent tools in healthcare, including ophthalmology. This systematic review offers a twofold contribution: it summarises current trends in ophthalmology-related LLM research and projects future directions for this burgeoning field., Methods: We systematically searched across various databases (PubMed, Europe PMC, Scopus and Web of Science) for articles related to LLM use in ophthalmology, published between 1 January 2022 and 31 July 2023. Selected articles were summarised, and categorised by type (editorial, commentary, original research, etc) and their research focus (eg, evaluating ChatGPT's performance in ophthalmology examinations or clinical tasks)., Findings: We identified 32 articles meeting our criteria, published between January and July 2023, with a peak in June (n=12). Most were original research evaluating LLMs' proficiency in clinically related tasks (n=9). Studies demonstrated that ChatGPT-4.0 outperformed its predecessor, ChatGPT-3.5, in ophthalmology exams. Furthermore, ChatGPT excelled in constructing discharge notes (n=2), evaluating diagnoses (n=2) and answering general medical queries (n=6). However, it struggled with generating scientific articles or abstracts (n=3) and answering specific subdomain questions, especially those regarding specific treatment options (n=2). ChatGPT's performance relative to other LLMs (Google's Bard, Microsoft's Bing) varied by study design. Ethical concerns such as data hallucination (n=27), authorship (n=5) and data privacy (n=2) were frequently cited., Interpretation: While LLMs hold transformative potential for healthcare and ophthalmology, concerns over accountability, accuracy and data security remain. Future research should focus on application programming interface integration, comparative assessments of popular LLMs, their ability to interpret image-based data and the establishment of standardised evaluation frameworks., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)

Purpose: The objective of this paper is to shed light on the current landscape of genotyping practices, phenotyping practices and availability of essential vision rehabilitation management for inherited retinal diseases (IRD) in the Asia-Pacific (APAC) Region., Methods: The 62-item questionnaire was distributed electronically via email. The questions covered five domains: (1) structure of the IRD service and registry/database; (2) genotyping practices; (3) genetic counselling; (4) deep phenotyping practices; (5) low-vision rehabilitation services., Results: The survey was completed by 36 of 45 centres in twelve countries and regions in APAC. Among these centres, 42 % reported managing more than 1000 patients. Notably, 39 % of centres lack an IRD database or registry, and 44 % of centres have tested less than one-quarter of their IRD patients. The majority of centres (67 %) do not have genetic counsellors. While there was consistency in the imaging-based investigations, there was marked heterogeneity for functional testing using electrophysiology and formal perimetry. Only 34 % of centres confirmed the availability of access to low-vision assistive devices., Conclusions: This study reveals several critical gaps in managing IRDs in the APAC region. These include the lack of IRD database/registry in one-third of centres, a substantial proportion of patients remaining genetically undiagnosed, and limited availability of genetic counsellors. The findings also underscore a need to harmonise investigations for evaluating retinal function and identify areas for improvement in the provision of low-vision rehabilitation services., (Copyright © 2024 The Authors. Published by Elsevier Inc. All rights reserved.)

Artificial intelligence (AI) use in diabetes care is increasingly being explored to personalise care for people with diabetes and adapt treatments for complex presentations. However, the rapid advancement of AI also introduces challenges such as potential biases, ethical considerations, and implementation challenges in ensuring that its deployment is equitable. Ensuring inclusive and ethical developments of AI technology can empower both health-care providers and people with diabetes in managing the condition. In this Review, we explore and summarise the current and future prospects of AI across the diabetes care continuum, from enhancing screening and diagnosis to optimising treatment and predicting and managing complications., Competing Interests: Declaration of interests NS is the director of London Safety and Training Solutions. HL is supported by the National Key Research and Development Program of China (2022YFA1004804), the Excellent Young Scientists Fund of the National Natural Science Foundation of China (82022012), and the innovative research team of high-level local universities in Shanghai (SHSMU-ZDCX20212700). Y-CT is supported by the National Medical Research Council of Singapore (NMRC/MOH/ HCSAINV21nov-0001)., (Copyright © 2024 Published by Elsevier Ltd. All rights reserved, including those for text and data mining, AI training, and similar technologies.)

Background: Deep learning (DL) is promising to detect glaucoma. However, patients' privacy and data security are major concerns when pooling all data for model development. We developed a privacy-preserving DL model using the federated learning (FL) paradigm to detect glaucoma from optical coherence tomography (OCT) images., Methods: This is a multicentre study. The FL paradigm consisted of a 'central server' and seven eye centres in Hong Kong, the USA and Singapore. Each centre first trained a model locally with its own OCT optic disc volumetric dataset and then uploaded its model parameters to the central server. The central server used FedProx algorithm to aggregate all centres' model parameters. Subsequently, the aggregated parameters are redistributed to each centre for its local model optimisation. We experimented with three three-dimensional (3D) networks to evaluate the stabilities of the FL paradigm. Lastly, we tested the FL model on two prospectively collected unseen datasets., Results: We used 9326 volumetric OCT scans from 2785 subjects. The FL model performed consistently well with different networks in 7 centres (accuracies 78.3%-98.5%, 75.9%-97.0%, and 78.3%-97.5%, respectively) and stably in the 2 unseen datasets (accuracies 84.8%-87.7%, 81.3%-84.8%, and 86.0%-87.8%, respectively). The FL model achieved non-inferior performance in classifying glaucoma compared with the traditional model and significantly outperformed the individual models., Conclusion: The 3D FL model could leverage all the datasets and achieve generalisable performance, without data exchange across centres. This study demonstrated an OCT-based FL paradigm for glaucoma identification with ensured patient privacy and data security, charting another course toward the real-world transition of artificial intelligence in ophthalmology., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)

Background: Patients requiring coronary artery bypass grafting (CABG) have multiple co-morbidities which need to be considered in totality when determining surgical risks. The objective of this study is to evaluate short-term and long-term mortality rates of CABG surgery, as well as to identify the most significant risk factors for mortality after isolated CABG., Methods: All patients with complete dataset who underwent isolated CABG between January 2008 and December 2017 were included. Univariate and multivariate Cox regression was performed to determine the risk factors for all-cause mortality. Classification and regression tree analysis was performed to identify the relative importance of these risk factors., Results: 3,573 patients were included in the study. Overall mortality rate was 25.7%. In-hospital mortality rate was 1.62% overall. 30-day, 1-year, 5-year, 10-year and 14.5-year mortality rates were 1.46%, 2.94%, 9.89%, 22.79% and 36.30% respectively. Factors associated with death after adjustment for other risk factors were older age, lower body mass index (BMI), hypertension, diabetes mellitus, chronic obstructive pulmonary disease, pre-operative renal failure on dialysis, higher last pre-operative creatinine level, lower estimated glomerular filtration rate (eGFR), heart failure, lower left ventricular ejection fraction and New York Heart Association class II, III and IV. Additionally, female gender and logistic EuroSCORE were associated with death on univariate Cox analysis, but not associated with death after adjustment with multivariate Cox analysis. Using CART analysis, the strongest predictor of mortality was pre-operative eGFR < 46.9, followed by logistic EuroSCORE ≥ 2.4., Conclusion: Poorer renal function, quantified by a lower eGFR, is the best predictor of post-CABG mortality. Amongst other risk factors, logistic EuroSCORE, age, diabetes and BMI had a relatively greater impact on mortality. Patients with chronic kidney disease stage 3B and above are at highest risk for mortality. We hope these findings heighten awareness to optimise current medical therapy in preserving renal function upon diagnosis of any atherosclerotic disease and risk factors contributing to coronary artery disease., (© 2024. The Author(s).)

Objective: The aim of this study was to evaluate the accuracy, comprehensiveness, and safety of a publicly available large language model (LLM)-ChatGPT in the sub-domain of glaucoma., Design: Evaluation of diagnostic test or technology., Subjects Participants And/or Controls: We seek to evaluate the responses of an artificial intelligence chatbot ChatGPT (version GPT-3.5, OpenAI)., Methods Intervention or Testing: We curated 24 clinically relevant questions in the domain of glaucoma. The questions spanned four categories: pertaining to diagnosis, treatment, surgeries, and ocular emergencies. Each question was posed to the LLM and the responses obtained were graded by an expert grader panel of three glaucoma specialists with combined experience of more than 30 years in the field. For responses which performed poorly, the LLM was further prompted to self-correct. The subsequent responses were then re-evaluated by the expert panel., Main Outcome Measures: Accuracy, comprehensiveness, and safety of the responses of a public domain LLM., Results: There were a total of 24 questions and three expert graders with a total number of responses of n  = 72. The scores were ranked from 1 to 4, where 4 represents the best score with a complete and accurate response. The mean score of the expert panel was 3.29 with a standard deviation of 0.484. Out of the 24 question-response pairs, seven (29.2%) of them had a mean inter-grader score of 3 or less. The mean score of the original seven question-response pairs was 2.96 which rose to 3.58 after an opportunity to self-correct (z-score - 3.27, p  = 0.001, Mann-Whitney U). The seven out of 24 question-response pairs which performed poorly were given a chance to self-correct. After self-correction, the proportion of responses obtaining a full score increased from 22/72 (30.6%) to 12/21 (57.1%), ( p  = 0.026, χ 2 test)., Conclusion: LLMs show great promise in the realm of glaucoma with additional capabilities of self-correction. The application of LLMs in glaucoma is still in its infancy, and still requires further research and validation., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision., (Copyright © 2024 Tan, Tham, Koh, Loon, Aquino, Lun, Cheng, Ngiam and Tan.)

Background: Uncorrected refractive error is a major cause of vision impairment worldwide and its increasing prevalent necessitates effective screening and management strategies. Meanwhile, deep learning, a subset of Artificial Intelligence, has significantly advanced ophthalmological diagnostics by automating tasks that required extensive clinical expertise. Although recent studies have investigated the use of deep learning models for refractive power detection through various imaging techniques, a comprehensive systematic review on this topic is has yet be done. This review aims to summarise and evaluate the performance of ocular image-based deep learning models in predicting refractive errors., Main Text: We search on three databases (PubMed, Scopus, Web of Science) up till June 2023, focusing on deep learning applications in detecting refractive error from ocular images. We included studies that had reported refractive error outcomes, regardless of publication years. We systematically extracted and evaluated the continuous outcomes (sphere, SE, cylinder) and categorical outcomes (myopia), ground truth measurements, ocular imaging modalities, deep learning models, and performance metrics, adhering to PRISMA guidelines. Nine studies were identified and categorised into three groups: retinal photo-based (n ​= ​5), OCT-based (n ​= ​1), and external ocular photo-based (n ​= ​3).For high myopia prediction, retinal photo-based models achieved AUC between 0.91 and 0.98, sensitivity levels between 85.10% and 97.80%, and specificity levels between 76.40% and 94.50%. For continuous prediction, retinal photo-based models reported MAE ranging from 0.31D to 2.19D, and R 2 between 0.05 and 0.96. The OCT-based model achieved an AUC of 0.79-0.81, sensitivity of 82.30% and 87.20% and specificity of 61.70%-68.90%. For external ocular photo-based models, the AUC ranged from 0.91 to 0.99, sensitivity of 81.13%-84.00% and specificity of 74.00%-86.42%, MAE ranges from 0.07D to 0.18D and accuracy ranges from 81.60% to 96.70%. The reported papers collectively showed promising performances, in particular the retinal photo-based and external eye photo -based DL models., Conclusions: The integration of deep learning model and ocular imaging for refractive error detection appear promising. However, their real-world clinical utility in current screening workflow have yet been evaluated and would require thoughtful consideration in design and implementation., Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (© 2024 The Authors.)

Purpose: There is a scarcity of literature focusing on sleep's impact on myopia in children despite an epidemic rise of myopia among the age group and the importance of early prevention. As such, this systematic review-meta-analysis aims to evaluate the association between various aspects of sleep and myopia in children and adolescents aged 0-19 years., Methods: We searched PubMed, EMBASE, and Cochrane Library on 08/12/2022 for studies reporting sleep in relation to myopia among children and adolescents. Myopia was defined as spherical equivalent refraction < -0.5 diopter. The primary outcome was the relationship between sleep duration and myopia prevalence. Secondary outcomes include the effect of sleep quality, bedtime, and waketime on myopia prevalence, incidence, and progression. Odds ratio (OR) was estimated with a 95% confidence interval (95% CI)., Results: Eighteen studies (49,277 participants) were included in the review, and six studies (14,116 participants) were included in the meta-analysis for the primary outcome. There was no significant correlation between sleep and myopia prevalence (OR = 0.905, 95% CI = 0.782 to 1.047). Some studies suggested that better sleep quality (2 of 6 studies), earlier bedtime (3 of 5 studies), and later waketimes (2 of 3 studies) had protective effects on myopia., Conclusion: Sleep duration did not affect myopia prevalence in children, while other aspects of sleep had plausible but inconclusive impacts on myopia development and progression. More research with diverse populations and standardized methods of reporting is needed., (© 2023. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.)

The emergence of generative artificial intelligence (AI) has revolutionized various fields. In ophthalmology, generative AI has the potential to enhance efficiency, accuracy, personalization and innovation in clinical practice and medical research, through processing data, streamlining medical documentation, facilitating patient-doctor communication, aiding in clinical decision-making, and simulating clinical trials. This review focuses on the development and integration of generative AI models into clinical workflows and scientific research of ophthalmology. It outlines the need for development of a standard framework for comprehensive assessments, robust evidence, and exploration of the potential of multimodal capabilities and intelligent agents. Additionally, the review addresses the risks in AI model development and application in clinical service and research of ophthalmology, including data privacy, data bias, adaptation friction, over interdependence, and job replacement, based on which we summarized a risk management framework to mitigate these concerns. This review highlights the transformative potential of generative AI in enhancing patient care, improving operational efficiency in the clinical service and research in ophthalmology. It also advocates for a balanced approach to its adoption., Competing Interests: Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: co-author is the Editor-in-Chief of the Asia-Pacific Journal of Ophthalmology (APJO) - Dennis Shun-chiu Lam. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 The Authors. Published by Elsevier Inc. All rights reserved.)

Aims: To determine axial length (AL) elongation profiles in children aged 3-6 years in an Asian population., Methods: Eligible subjects were recruited from the Growing Up in Singapore Towards Healthy Outcomes birth cohort. AL measurement was performed using IOLMaster (Carl Zeiss Meditec, Jena, Germany) at 3 and 6 years. Anthropometric measurements at birth, cycloplegic refraction at 3 and 6 years, questionnaires on the children's behavioural habits at 2 years and parental spherical equivalent refraction were performed. Multivariable linear regression model with generalised estimating equation was performed to determine factors associated with AL elongation., Results: 273 eyes of 194 children were included. The mean AL increased from 21.72±0.59 mm at 3 years to 22.52±0.66 mm at 6 years (p<0.001). Myopic eyes at 6 years had greater AL elongation (1.02±0.34 mm) compared with emmetropic eyes (0.85±0.25 mm, p=0.008) and hyperopic eyes (0.74±0.16 mm, p<0.001). The 95th percentile limit of AL elongation was 1.59 mm in myopes, 1.34 mm in emmetropes and 1.00 mm in hyperopes. Greater birth weight (per 100 g, β=0.010, p=0.02) was significantly associated with greater AL elongation from 3 to 6 years, while parental and other behavioural factors assessed at 2 years were not (all p≥0.08)., Conclusion: In this preschool cohort, AL elongates at an average length of 0.80 mm from 3 to 6 years, with myopes demonstrating the greatest elongation. The differences in 95th percentile limits for AL elongation between myopes, emmetropes and hyperopes can be valuable information in identifying myopia development in preschool children., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)

Purpose: Chronic kidney disease (CKD) may elevate susceptibility to age-related macular degeneration (AMD) because of shared risk factors, pathogenic mechanisms, and genetic polymorphisms. Given the inconclusive findings in prior studies, we investigated this association using extensive datasets in the Asian Eye Epidemiology Consortium., Design: Cross-sectional study., Participants: Fifty-one thousand two hundred fifty-three participants from 10 distinct population-based Asian studies., Methods: Age-related macular degeneration was defined using the Wisconsin Age-Related Maculopathy Grading System, the International Age-Related Maculopathy Epidemiological Study Group Classification, or the Beckman Clinical Classification. Chronic kidney disease was defined as estimated glomerular filtration rate (eGFR) of less than 60 ml/min per 1.73 m 2 . A pooled analysis using individual-level participant data was performed to examine the associations between CKD and eGFR with AMD (early and late), adjusting for age, sex, hypertension, diabetes, body mass index, smoking status, total cholesterol, and study groups., Main Outcome Measures: Odds ratio (OR) of early and late AMD., Results: Among 51 253 participants (mean age, 54.1 ± 14.5 years), 5079 had CKD (9.9%). The prevalence of early AMD was 9.0%, and that of late AMD was 0.71%. After adjusting for confounders, individuals with CKD were associated with higher odds of late AMD (OR, 1.46; 95% confidence interval [CI], 1.11-1.93; P = 0.008). Similarly, poorer kidney function (per 10-unit eGFR decrease) was associated with late AMD (OR, 1.12; 95% CI, 1.05-1.19; P = 0.001). Nevertheless, CKD and eGFR were not associated significantly with early AMD (all P ≥ 0.149)., Conclusions: Pooled analysis from 10 distinct Asian population-based studies revealed that CKD and compromised kidney function are associated significantly with late AMD. This finding further underscores the importance of ocular examinations in patients with CKD., Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (Copyright © 2024 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.)

Objective: Vision transformers (ViTs) have shown promising performance in various classification tasks previously dominated by convolutional neural networks (CNNs). However, the performance of ViTs in referable diabetic retinopathy (DR) detection is relatively underexplored. In this study, using retinal photographs, we evaluated the comparative performances of ViTs and CNNs on detection of referable DR., Design: Retrospective study., Participants: A total of 48 269 retinal images from the open-source Kaggle DR detection dataset, the Messidor-1 dataset and the Singapore Epidemiology of Eye Diseases (SEED) study were included., Methods: Using 41 614 retinal photographs from the Kaggle dataset, we developed 5 CNN (Visual Geometry Group 19, ResNet50, InceptionV3, DenseNet201, and EfficientNetV2S) and 4 ViTs models (VAN&#95;small, CrossViT&#95;small, ViT&#95;small, and Hierarchical Vision transformer using Shifted Windows [SWIN]&#95;tiny) for the detection of referable DR. We defined the presence of referable DR as eyes with moderate or worse DR. The comparative performance of all 9 models was evaluated in the Kaggle internal test dataset (with 1045 study eyes), and in 2 external test sets, the SEED study (5455 study eyes) and the Messidor-1 (1200 study eyes)., Main Outcome Measures: Area under operating characteristics curve (AUC), specificity, and sensitivity., Results: Among all models, the SWIN transformer displayed the highest AUC of 95.7% on the internal test set, significantly outperforming the CNN models (all P  < 0.001). The same observation was confirmed in the external test sets, with the SWIN transformer achieving AUC of 97.3% in SEED and 96.3% in Messidor-1. When specificity level was fixed at 80% for the internal test, the SWIN transformer achieved the highest sensitivity of 94.4%, significantly better than all the CNN models (sensitivity levels ranging between 76.3% and 83.8%; all P  < 0.001). This trend was also consistently observed in both external test sets., Conclusions: Our findings demonstrate that ViTs provide superior performance over CNNs in detecting referable DR from retinal photographs. These results point to the potential of utilizing ViT models to improve and optimize retinal photo-based deep learning for referable DR detection., Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (© 2024 by the American Academy of Ophthalmology.)

Background: Artificial intelligence (AI) that utilizes deep learning (DL) has potential for systemic disease prediction using retinal imaging. The retina's unique features enable non-invasive visualization of the central nervous system and microvascular circulation, aiding early detection and personalized treatment plans for personalized care. This review explores the value of retinal assessment, AI-based retinal biomarkers, and the importance of longitudinal prediction models in personalized care., Main Text: This narrative review extensively surveys the literature for relevant studies in PubMed and Google Scholar, investigating the application of AI-based retina biomarkers in predicting systemic diseases using retinal fundus photography. The study settings, sample sizes, utilized AI models and corresponding results were extracted and analysed. This review highlights the substantial potential of AI-based retinal biomarkers in predicting neurodegenerative, cardiovascular, and chronic kidney diseases. Notably, DL algorithms have demonstrated effectiveness in identifying retinal image features associated with cognitive decline, dementia, Parkinson's disease, and cardiovascular risk factors. Furthermore, longitudinal prediction models leveraging retinal images have shown potential in continuous disease risk assessment and early detection. AI-based retinal biomarkers are non-invasive, accurate, and efficient for disease forecasting and personalized care., Conclusion: AI-based retinal imaging hold promise in transforming primary care and systemic disease management. Together, the retina's unique features and the power of AI enable early detection, risk stratification, and help revolutionizing disease management plans. However, to fully realize the potential of AI in this domain, further research and validation in real-world settings are essential., (© 2024. The Author(s).)

We set out to estimate the international incidence of rhegmatogenous retinal detachment (RRD) and to evaluate its temporal trend over time. There is a lack of robust estimates on the worldwide incidence and trend for RRD, a major cause of acute vision loss. We conducted a systematic review of RRD incidence. The electronic databases PubMed, Scopus, and Thomson Reuters' Web of Science were searched from inception through 2nd June 2022. Random-effects meta-analysis model with logit transformation was performed to obtain pooled annual incidence estimates of RRD. Pooled analysis was performed to evaluate the temporal trend of RRD incidence of the 20,958 records identified from the database searches; 33 studies from 21 countries were included for analysis (274,836 cases of RRD in 273,977 persons). Three of the 6 global regions as defined by WHO had studies that met the inclusion and exclusion criteria of the study. The annual international incidence of RRD was estimated to be 12.17 (95% confidence interval [CI] 10.51-14.09) per 100,000 population; with an increasing temporal trend of RRD at 5.4 per 100,000 per decade (p 0.001) from 1997 to 2019. Amongst world regions, the RRD incidence was highest in Europe (14.52 [95% CI 11.79 - 17.88] per 100,000 population), followed by Western Pacific (10.55 [95% CI 8.71-12.75] per 100,000 population) and Regions of Americas (8.95 [95% CI 6.73-11.92] per 100,000 population). About one in 10,000 persons develop RRD each year. There is evidence of increasing trend for RRD incidence over time, with possibly doubling of the current incidence rate within the next 2 decades., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024. Published by Elsevier Inc.)

Objective: Our objective was to determine the effects of lipids and complement proteins on early and intermediate age-related macular degeneration (AMD) stages using machine learning models by integrating metabolomics and proteomic data., Design: Nested case-control study., Subjects and Controls: The analyses were performed in a subset of the Singapore Indian Chinese Cohort (SICC) Eye Study. Among the 6753 participants, we randomly selected 155 Indian and 155 Chinese cases of AMD and matched them with 310 controls on age, sex, and ethnicity., Methods: We measured 35 complement proteins and 56 lipids using mass spectrometry and nuclear magnetic resonance, respectively. We first selected the most contributing lipids and complement proteins to early and intermediate AMD using random forest models. Then, we estimated their effects using a multinomial model adjusted for potential confounders., Main Outcome Measures: Age-related macular degeneration was classified using the Beckman classification system., Results: Among the 310 individuals with AMD, 166 (53.5%) had early AMD, and 144 (46.5%) had intermediate AMD. First, high-density lipoprotein (HDL) particle diameter was positively associated with both early and intermediate AMD (odds ratio [OR] early  = 1.69; 95% confidence interval [CI],1.11-2.55 and OR intermediate  = 1.72; 95% CI, 1.11-2.66 per 1-standard deviation increase in HDL diameter). Second, complement protein 2 (C2), complement C1 inhibitor (IC1), complement protein 6 (C6), complement protein 1QC (C1QC) and complement factor H-related protein 1 (FHR1), were associated with AMD. C2 was positively associated with both early and intermediate AMD (OR early  = 1.58; 95% CI, 1.08-2.30 and OR intermediate  = 1.56; 95% CI, 1.04-2.34). C6 was positively (OR early  = 1.41; 95% CI, 1.03-1.93) associated with early AMD. However, IC1 was negatively associated with early AMD (OR early  = 0.62; 95% CI, 0.38-0.99), whereas C1QC (OR intermediate  = 0.63; 95% CI, 0.42-0.93) and FHR1 (OR intermediate  = 0.73; 95% CI, 0.54-0.98) were both negatively associated with intermediate AMD., Conclusions: Although both HDL diameter and C2 levels show associations with both early and intermediate AMD, dysregulations of IC1, C6, C1QC, and FHR1 are only observed at specific stages of AMD. These findings underscore the complexity of complement system dysregulation in AMD, which appears to vary depending on the disease severity., Financial Disclosures: The authors have no proprietary or commercial interest in any materials discussed in this article., (© 2024 by the American Academy of Ophthalmology.)

Background: The convergence of smartphone technology and artificial intelligence (AI) has revolutionized the landscape of ophthalmic care, offering unprecedented opportunities for diagnosis, monitoring, and management of ocular conditions. Nevertheless, there is a lack of systematic studies on discussing the integration of smartphone and AI in this field., Main Text: This review includes 52 studies, and explores the integration of smartphones and AI in ophthalmology, delineating its collective impact on screening methodologies, disease detection, telemedicine initiatives, and patient management. The collective findings from the curated studies indicate promising performance of the smartphone-based AI screening for various ocular diseases which encompass major retinal diseases, glaucoma, cataract, visual impairment in children and ocular surface diseases. Moreover, the utilization of smartphone-based imaging modalities, coupled with AI algorithms, is able to provide timely, efficient and cost-effective screening for ocular pathologies. This modality can also facilitate patient self-monitoring, remote patient monitoring and enhancing accessibility to eye care services, particularly in underserved regions. Challenges involving data privacy, algorithm validation, regulatory frameworks and issues of trust are still need to be addressed. Furthermore, evaluation on real-world implementation is imperative as well, and real-world prospective studies are currently lacking., Conclusions: Smartphone ocular imaging merged with AI enables earlier, precise diagnoses, personalized treatments, and enhanced service accessibility in eye care. Collaboration is crucial to navigate ethical and data security challenges while responsibly leveraging these innovations, promising a potential revolution in care access and global eye health equity., Competing Interests: The authors affirm that there are no apparent financial interests or personal relationships that might have influenced the reported work in this paper., (© 2024 Published by Elsevier Inc. on behalf of Zhejiang University Press.)

Objective: To determine the incidence and risk factors for primary open-angle glaucoma (POAG) and ocular hypertension (OHT) in a multiethnic Asian population., Design: Population-based cohort study., Participants: The Singapore Epidemiology of Eye Diseases study included 10 033 participants in the baseline examination between 2004 and 2011. Of those, 6762 (response rate = 78.8%) participated in the 6-year follow-up visit between 2011 and 2017., Methods: Standardized examination and investigations were performed, including slit lamp biomicroscopy, intraocular pressure (IOP) measurement, pachymetry, gonioscopy, optic disc examination and static automated perimetry. Glaucoma was defined according to a combination of clinical evaluation, ocular imaging (fundus photo, visual field, and OCT) and criteria given by International Society of Geographical and Epidemiological Ophthalmology. OHT was defined on the basis of elevated IOP over the upper limit of normal; i.e., 20.4 mmHg, 21.5 mmHg, and 22.6 mmHg for the Chinese, Indian, and Malay cohort respectively, without glaucomatous optic disc change., Main Outcome Measures: Incidence of POAG, OHT, and OHT progression., Results: The overall 6-year age-adjusted incidences of POAG and OHT were 1.31% (95% confidence interval [CI], 1.04-1.62) and 0.47% (95% CI, 0.30-0.70). The rate of progression of baseline OHT to POAG at 6 years was 5.32%. Primary open-angle glaucoma incidence was similar (1.37%) in Chinese and Indians and lower (0.80%) in Malays. Malays had higher incidence (0.79%) of OHT than Indians (0.38%) and Chinese (0.37%). Baseline parameters associated with higher risk of POAG were older age (per decade: odds ratio [OR], 1.90; 95% CI, 1.54-2.35; P < 0.001), higher baseline IOP (per mmHg: OR, 1.20; 95% CI, 1.12-1.29; P < 0.001) and longer axial length (per mm: OR, 1.22; 95% CI, 1.07-1.40, P = 0.004)., Conclusion: Six-year incidence of POAG was 1.31% in a multiethnic Asian population. Older age, higher IOP, and longer axial length were associated with higher risk of POAG. These findings can help in future projections and guide public healthcare policy decisions for screening at-risk individuals., Financial Disclosure(s): The authors have no proprietary or commercial interest in any materials discussed in this article., (Copyright © 2023 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.)

Abstract: With the rise of generative artificial intelligence (AI) and AI-powered chatbots, the landscape of medicine and healthcare is on the brink of significant transformation. This perspective delves into the prospective influence of AI on medical education, residency training and the continuing education of attending physicians or consultants. We begin by highlighting the constraints of the current education model, challenges in limited faculty, uniformity amidst burgeoning medical knowledge and the limitations in 'traditional' linear knowledge acquisition. We introduce 'AI-assisted' and 'AI-integrated' paradigms for medical education and physician training, targeting a more universal, accessible, high-quality and interconnected educational journey. We differentiate between essential knowledge for all physicians, specialised insights for clinician-scientists and mastery-level proficiency for clinician-computer scientists. With the transformative potential of AI in healthcare and service delivery, it is poised to reshape the pedagogy of medical education and residency training., (Copyright © 2024 Copyright: © 2024 Singapore Medical Journal.)

Objectives: To provide balanced consideration of the opportunities and challenges associated with integrating Large Language Models (LLMs) throughout the medical school continuum., Process: Narrative review of published literature contextualized by current reports of LLM application in medical education., Conclusions: LLMs like OpenAI's ChatGPT can potentially revolutionize traditional teaching methodologies. LLMs offer several potential advantages to students, including direct access to vast information, facilitation of personalized learning experiences, and enhancement of clinical skills development. For faculty and instructors, LLMs can facilitate innovative approaches to teaching complex medical concepts and fostering student engagement. Notable challenges of LLMs integration include the risk of fostering academic misconduct, inadvertent overreliance on AI, potential dilution of critical thinking skills, concerns regarding the accuracy and reliability of LLM-generated content, and the possible implications on teaching staff., (© The Author(s) 2024. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For permissions, please email: journals.permissions@oup.com.)

We described a challenge named "DRAC - Diabetic Retinopathy Analysis Challenge" in conjunction with the 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2022). Within this challenge, we provided the DRAC datset, an ultra-wide optical coherence tomography angiography (UW-OCTA) dataset (1,103 images), addressing three primary clinical tasks: diabetic retinopathy (DR) lesion segmentation, image quality assessment, and DR grading. The scientific community responded positively to the challenge, with 11, 12, and 13 teams submitting different solutions for these three tasks, respectively. This paper presents a concise summary and analysis of the top-performing solutions and results across all challenge tasks. These solutions could provide practical guidance for developing accurate classification and segmentation models for image quality assessment and DR diagnosis using UW-OCTA images, potentially improving the diagnostic capabilities of healthcare professionals. The dataset has been released to support the development of computer-aided diagnostic systems for DR evaluation., Competing Interests: The authors declare no competing interests., (© 2024 The Author(s).)

Diabetic retinopathy (DR) is the leading cause of preventable blindness worldwide. The risk of DR progression is highly variable among different individuals, making it difficult to predict risk and personalize screening intervals. We developed and validated a deep learning system (DeepDR Plus) to predict time to DR progression within 5 years solely from fundus images. First, we used 717,308 fundus images from 179,327 participants with diabetes to pretrain the system. Subsequently, we trained and validated the system with a multiethnic dataset comprising 118,868 images from 29,868 participants with diabetes. For predicting time to DR progression, the system achieved concordance indexes of 0.754-0.846 and integrated Brier scores of 0.153-0.241 for all times up to 5 years. Furthermore, we validated the system in real-world cohorts of participants with diabetes. The integration with clinical workflow could potentially extend the mean screening interval from 12 months to 31.97 months, and the percentage of participants recommended to be screened at 1-5 years was 30.62%, 20.00%, 19.63%, 11.85% and 17.89%, respectively, while delayed detection of progression to vision-threatening DR was 0.18%. Altogether, the DeepDR Plus system could predict individualized risk and time to DR progression over 5 years, potentially allowing personalized screening intervals., (© 2024. The Author(s).)

Cardiovascular disease (CVD) remains the leading cause of death worldwide. Assessing of CVD risk plays an essential role in identifying individuals at higher risk and enables the implementation of targeted intervention strategies, leading to improved CVD prevalence reduction and patient survival rates. The ocular vasculature, particularly the retinal vasculature, has emerged as a potential means for CVD risk stratification due to its anatomical similarities and physiological characteristics shared with other vital organs, such as the brain and heart. The integration of artificial intelligence (AI) into ocular imaging has the potential to overcome limitations associated with traditional semi-automated image analysis, including inefficiency and manual measurement errors. Furthermore, AI techniques may uncover novel and subtle features that contribute to the identification of ocular biomarkers associated with CVD. This review provides a comprehensive overview of advancements made in AI-based ocular image analysis for predicting CVD, including the prediction of CVD risk factors, the replacement of traditional CVD biomarkers (e.g., CT-scan measured coronary artery calcium score), and the prediction of symptomatic CVD events. The review covers a range of ocular imaging modalities, including colour fundus photography, optical coherence tomography, and optical coherence tomography angiography, and other types of images like external eye images. Additionally, the review addresses the current limitations of AI research in this field and discusses the challenges associated with translating AI algorithms into clinical practice., (© 2023. The Author(s), under exclusive licence to The Royal College of Ophthalmologists.)

X-chromosomal genetic variants are understudied but can yield valuable insights into sexually dimorphic human traits and diseases. We performed a sex-stratified cross-ancestry X-chromosome-wide association meta-analysis of seven kidney-related traits (n = 908,697), identifying 23 loci genome-wide significantly associated with two of the traits: 7 for uric acid and 16 for estimated glomerular filtration rate (eGFR), including four novel eGFR loci containing the functionally plausible prioritized genes ACSL4, CLDN2, TSPAN6 and the female-specific DRP2. Further, we identified five novel sex-interactions, comprising male-specific effects at FAM9B and AR/EDA2R, and three sex-differential findings with larger genetic effect sizes in males at DCAF12L1 and MST4 and larger effect sizes in females at HPRT1. All prioritized genes in loci showing significant sex-interactions were located next to androgen response elements (ARE). Five ARE genes showed sex-differential expressions. This study contributes new insights into sex-dimorphisms of kidney traits along with new prioritized gene targets for further molecular research., (© 2024. The Author(s).)

Purpose: There are major gaps in our knowledge of hereditary ocular conditions in the Asia-Pacific population, which comprises approximately 60% of the world's population. Therefore, a concerted regional effort is urgently needed to close this critical knowledge gap and apply precision medicine technology to improve the quality of lives of these patients in the Asia-Pacific region., Design: Multi-national, multi-center collaborative network., Methods: The Research Standing Committee of the Asia-Pacific Academy of Ophthalmology and the Asia-Pacific Society of Eye Genetics fostered this research collaboration, which brings together renowned institutions and experts for inherited eye diseases in the Asia-Pacific region. The immediate priority of the network will be inherited retinal diseases (IRDs), where there is a lack of detailed characterization of these conditions and in the number of established registries., Results: The network comprises 55 members from 35 centers, spanning 12 countries and regions, including Australia, China, India, Indonesia, Japan, South Korea, Malaysia, Nepal, Philippines, Singapore, Taiwan, and Thailand. The steering committee comprises ophthalmologists with experience in consortia for eye diseases in the Asia-Pacific region, leading ophthalmologists and vision scientists in the field of IRDs internationally, and ophthalmic geneticists., Conclusions: The Asia Pacific Inherited Eye Disease (APIED) network aims to (1) improve genotyping capabilities and expertise to increase early and accurate genetic diagnosis of IRDs, (2) harmonise deep phenotyping practices and utilization of ontological terms, and (3) establish high-quality, multi-user, federated disease registries that will facilitate patient care, genetic counseling, and research of IRDs regionally and internationally., Competing Interests: Declaration of Competing Interest The authors have no conflicts of interest to disclose., (Copyright © 2023 The Authors. Published by Elsevier Inc. All rights reserved.)