KBKunho Bae

Kunho Bae

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Kunho Bae has a total of 75 co-authors. Here are the top co-authors by number of shared publications:

  • Chang Ki Yoon

    Department of Ophthalmology, Seoul National University Hospital, Seoul, Republic of Korea

    6 shared publications
  • Un Chul Park

    Department of Ophthalmology, Seoul National University Hospital, Seoul, Republic of Korea

    6 shared publications
  • Ju-Yeun Lee

    Department of Ophthalmology, Myongji Hospital, Hanyang University College of Medicine, Goyang, Republic of Korea

    5 shared publications
  • Eun Kyoung Lee

    Department of Ophthalmology, Seoul National University Hospital, Seoul, Republic of Korea

    4 shared publications
  • Eun Kyoung Lee

    Department of Ophthalmology, Seoul National University College of Medicine, Seoul National University Hospital, Seoul, Korea

    4 shared publications
  • Ju-Yeun Lee

    Department of Ophthalmology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, United States of America

    4 shared publications

Latest Publications

  • Open Access

    Deep learning–derived retinal age gap and its associations with lifestyle, systemic, and ocular health in a health screening cohort

    by Boa Jang, Richul Oh, Tae-Hoon Lee, Chang Ki Yoon, Hyuk Jin Choi, Jinwook Choi, Young-Gon Kim, Kunho Bae

    Individuals of the same chronological age differ in biological aging, and scalable, noninvasive markers are needed. The deep learning–derived retinal age gap (RAG) is a promising measure of retinal aging, but its associations with real-world health determinants remain unclear. We developed a multi-task model to predict retinal age using 29,530 fundus images from 7535 participants in a health screening cohort and evaluated RAG in cohort A for lifestyle, socioeconomic, and systemic factors ( n = 5606) and cohort B for ocular diseases ( n = 1810). The bias-corrected multi-task model trained on mixed data achieved the best performance, with a mean absolute error of 2.656 years and a Pearson correlation of 0.921 in cohort A and 2.529 years and 0.938 in cohort B. Higher RAG was significantly associated with smoking (ex-smokers, β = +0.46 years; current smokers, β = +0.50 years) and with clinical diabetes (+2.52 years); both survived false discovery rate (FDR) and Bonferroni correction, and the diabetes association persisted across all sequential covariate-adjustment sets. Married participants had lower RAG ( β = −0.46 years), significant after FDR correction only. Hypertension and hyperlipidemia were not associated with RAG. In cohort B, RAG was significantly higher in eyes with age-related macular degeneration ( β = +0.60 years) and cataract ( β = +1.86 years) than in normal controls, both surviving corrections. RAG, an imaging-derived age-prediction residual, is therefore associated with lifestyle, systemic, and ocular health. Whether it reflects biological aging requires longitudinal validation against established aging biomarkers; at present, RAG suits population-level characterization better than individual-level risk stratification.

  • Open Access

    Evaluating an AI-assisted triage workflow for retinal diseases

    by Doohyun Park, Richul Oh, Jihyeon Baek, Kunho Bae

    The clinical value of artificial intelligence (AI) in fundus photography depends on workflow efficiency as well as diagnostic performance. We evaluated an AI-assisted negative-screening workflow for diabetic retinopathy (DR), retinal vein occlusion (RVO), and age-related macular degeneration (AMD) using 6,904 color fundus photographs independently graded by three retina specialists. In this workflow, AI-negative images were classified as negative without ophthalmologist review, whereas AI-positive images were referred for human interpretation. For each reader–disease pair, the reference standard was agreement between the other two readers; discordant cases were excluded (DR, 1.2%–2.3%; RVO, 0.3%–0.6%; AMD, 6.1%–11.6%). The workflow reduced direct ophthalmologist review to 7.3%–8.0% of images for DR, 11.7%–11.9% for RVO, and 13.5%–16.8% for AMD. For DR, sensitivity decreased significantly (0.9168 to 0.8838; difference, -0.0330; 95% confidence interval [CI], -0.0495 to -0.0165; p < 0.001), whereas specificity did not change significantly (0.9961 to 0.9983; p = 0.183). For RVO, sensitivity was unchanged (0.9807) and specificity did not change significantly (0.9985 to 0.9988; p = 0.320). For AMD, sensitivity decreased significantly (0.8570 to 0.8445; difference, -0.0124; 95% CI, -0.0215 to -0.0033; p = 0.008), whereas specificity did not change significantly (0.9685 to 0.9843; p = 0.324). In an exploratory image-level analysis for at least one target disease, review decreased to 27.7%–30.1%; sensitivity changed from 0.9122 to 0.9039 (difference, -0.0083; 95% CI, -0.0131 to -0.0035; p < 0.001) and specificity from 0.9659 to 0.9804 ( p = 0.334). AI-assisted negative-screening can reduce ophthalmologist workload across multiple retinal diseases, but with disease-specific safety trade-offs: because AI-negative images are not reviewed, reduced sensitivity for DR and AMD means a small proportion of true-positive cases would go undetected, potentially delaying diagnosis and treatment for sight-threatening disease. These results support disease-specific implementation strategies that balance workload reduction against missed positive cases.

  • Open Access

    Evaluating Artificial Intelligence as a First-Pass Reader in Fundus Photograph Screening: A Multireader Workflow Validation Study

    by Jihyeon Baek, Richul Oh, Doohyun Park, Kunho Bae

    Background/Objectives: Double reading with arbitration improves diagnostic reliability in fundus screening but requires repeated human interpretation. Whether artificial intelligence (AI) can serve as a first-pass decision source within this workflow, and whether this applies consistently across retinal diseases with differing inter-reader agreement, remains unclear. Methods: In this retrospective study, 6904 color fundus photographs from 2593 patients at a tertiary screening center were analyzed for age-related macular degeneration (AMD), diabetic retinopathy (DR), and retinal vein occlusion (RVO). Three retina specialists independently labeled each image, and an AI system provided binary classifications at a prespecified operating threshold targeting 0.99-sensitivity. In AI–human double reading, the AI and one reader independently interpreted each image, and a second reader arbitrated discordant cases; this was compared with conventional human–human double reading. Three reader combinations were evaluated per disease. Results: AI–human reading required 1.02–1.11 human reads per image versus 2.00–2.10 for human–human reading. For DR and RVO, human–human reading yielded sensitivities of 0.974 and 0.982 and specificities of 0.999 and 1.000, respectively. Across AI–human combinations, sensitivity and specificity did not differ significantly from human–human reading (all p ≥ 0.05; specificity differences ≤0.001). For AMD (human–human sensitivity 0.845, specificity 1.000), AI–human sensitivity varied: two combinations were higher (0.916 and 0.950; both p < 0.001) and one comparable (0.842; p = 0.742). AMD specificity remained ≥0.978. Conclusions: AI–human reading halved human reading volume without significant loss for DR and RVO; AMD varied by configuration. AI use within double-reading workflows should account for disease-specific inter-reader agreement, reader composition, and operating threshold. This was a single-center, retrospective study, prospective external validation in a multicenter setting is warranted before clinical implementation.

  • Open Access

    Behavioral and Structural Correlates of Axial Length in School-Aged Children: Baseline Findings from the Seoul Myopia Cohort Study

    by Ju-Yeun Lee, Chang Hwan Lee, Kyungsik Kim, Un Chul Park, Kunho Bae

    Digital device use has substantially altered children’s visual environments, yet the specific behaviors associated with axial elongation and early retinal remodeling remain unclear in highly urbanized settings. This study investigated biological and behavioral correlates of axial length (AL) and associated macular microstructural changes in preadolescent children. Participants underwent ophthalmic examinations and completed questionnaires assessing lifestyle and digital device use factors. Multivariable linear mixed-effects models identified factors associated with AL and macular thickness. The participant-level prevalence of myopia, defined as spherical equivalent ≤ −0.50 D, was 53.2% and increased from 37.1% in Grade 1 to a peak of 78.7% in Grade 4. Older age, a greater number of myopic parents, flatter corneal curvature, male sex, continuous smart device use exceeding 1 h per session, and viewing distance < 30 cm were independently associated with longer AL. Each 1 mm increase in AL was associated with significant subfoveal choroidal thinning and generalized extrafoveal macular thinning, whereas the central fovea remained relatively preserved. Pediatric AL was associated with hereditary susceptibility, specific digital viewing behaviors, and early extrafoveal retinal structural variation.

  • Open Access

    Impact of gestational diabetes on future diabetic retinopathy and microvascular complications: a nationwide cohort study

    by Sang Ha Lee, Kyung-Shin Lee, Ju-Yeun Lee, Seung Woo Choi, Eun Kyoung Lee, Chang Ki Yoon, Un Chul Park, Kyu Hyung Park, Kunho Bae

    Gestational diabetes mellitus (GDM) is a well-established risk factor for subsequent type 2 diabetes mellitus (T2DM). However, its long-term association with diabetic microvascular complications, particularly diabetic retinopathy (DR), and the joint role of polycystic ovary syndrome (PCOS) remain incompletely understood. Using the Korean National Health Insurance Service database, we conducted a nationwide retrospective cohort study including 406,309 women who had their index delivery between 2005 and 2012. Standard Cox proportional hazards models were used to evaluate the incidence of T2DM and systemic complications. For DR and vision-threatening DR (VTDR), time-varying covariate Cox models were applied in the full cohort, with incident T2DM treated as a time-dependent variable. Sensitivity analyses restricted to women who developed T2DM were also performed. Stratified analyses evaluated the joint and independent effects of coexisting PCOS. During follow-up, women with prior GDM had a higher risk of developing T2DM (adjusted hazard ratio [aHR] 3.52; 95% CI 3.26-3.80). In time-varying analyses, prior GDM was associated with an increased risk of vision-threatening DR (VTDR) (aHR 2.28; 95% CI 1.06-4.92), whereas overall DR did not show a consistent positive association across analysis. Women with both GDM and PCOS had the highest risk of T2DM and VTDR compared with those with neither condition (aHR 4.96 and 31.90, respectively). In contrast, although some macrovascular outcomes showed statistically significant associations, the overall magnitude of risk increase was modest. A history of GDM was associated with an increased long-term risk of severe diabetic retinal outcomes, particularly VTDR, after the onset of T2DM. The coexistence of PCOS is associated with a higher overall risk burden, with women having both conditions showing the highest risks. These findings support the need for individualized metabolic and ophthalmic surveillance in women with prior GDM, particularly those with coexisting PCOS.

  • Open Access

    Predictors of visual outcomes and recurrence after pars plana vitrectomy for uveitis: A 3-Year longitudinal cohort study

    by Jae Hyup Lee, Eun Kyoung Lee, Chang Ki Yoon, Un Chul Park, Kyu Hyung Park, Kunho Bae

    Uveitis is a chronic relapsing inflammatory disease that often requires long-term systemic immunomodulatory (IMD) therapy. Pars plana vitrectomy (PPV) is increasingly used as an adjunctive treatment for selected uveitis cases; however, robust long-term evidence remains limited. This retrospective cohort study evaluated 3-year outcomes in 159 uveitic eyes undergoing PPV and a reference group of 22 eyes receiving medical therapy alone to assess visual, anatomical, and systemic treatment burden changes. Following PPV, best-corrected visual acuity (BCVA) improved significantly and remained stable through 36 months (P < 0.001). Central macular thickness decreased from 326.8 to 288.4 μm (P < 0.001), and the proportion of eyes with vitreous haze < grade 1 increased from 24.5% to 90.0%. Systemic corticosteroid doses and the number of IMD agents were significantly reduced at 36 months (both P < 0.001), whereas these improvements were not observed in the reference group. Multivariable analyses identified worse baseline BCVA and higher preoperative intraocular pressure as predictors of poorer visual outcomes. Higher preoperative corticosteroid dose (> 15 mg/day) was associated with improved visual recovery but increased recurrence risk. These findings suggest that PPV is associated with durable visual and inflammatory improvements, and a reduction in systemic treatment burden in uveitic eyes.

  • Open Access

    Retinal Arteriosclerosis in a Large Health Screening Cohort: Associations with Systemic Vascular Comorbidities and Stroke in Young Adults

    by Kunho Bae, Ju-Yeun Lee, Hyuk Jin Choi

    Background: Routine fundus photography is widely accessible; however, its utility in stratifying systemic vascular risk in asymptomatic, general populations remains understudied. We utilized a large-scale health screening cohort in South Korea to evaluate the clinical validity of the retinal arteriosclerosis index (RAI) in a generally healthy population. Methods: We conducted a cross-sectional study of 74,608 adults who underwent routine health screening (2003–2010) at a tertiary center. Retinal arteriosclerosis was graded (0–4) by masked readers with a modified Scheie classification; a higher eye grade was defined as a person-level grade. High-grade RAI was prespecified as ≥2. Associations with systemic conditions (hypertension, type 2 diabetes, hyperlipidemia, metabolic syndrome, cardiovascular disease, and stroke) were examined by using multivariable logistic regression adjusted for demographic, lifestyle, and laboratory covariates; moreover, analyses were stratified by age and sex. Results: High-grade RAI was present in 4.5% of the participants and increased with age. After adjustment, high-grade RAI was associated with hypertension (OR, 2.97; 95% CI, 2.73–3.23), diabetes mellitus (OR, 1.35; 95% CI, 1.22–1.50), cardiovascular disease (OR, 1.46; 95% CI, 1.25–1.71), metabolic syndrome (OR, 1.63; 95% CI, 1.49–1.78), and stroke (OR, 1.98; 95% CI, 1.41–2.79) but not with hyperlipidemia. Sex-stratified analyses revealed broadly similar patterns, although high-grade RAI was associated with stroke in women and cardiovascular disease in men. Age-stratified analyses demonstrated consistent associations with hypertension, metabolic syndrome, and stroke across all age groups, with stronger effect sizes being observed in younger individuals. With respect to lifestyle factors, frequent alcohol consumption was associated with higher odds of high-grade RAI. Laboratory correlates included higher uric acid levels and lower red blood cell, albumin, and bilirubin levels (all p < 0.001). Conclusions: Fundus-defined arteriosclerotic changes were independently associated with multiple systemic vascular and metabolic conditions. An association with stroke in adults younger than 40 years of age was also observed, although this finding should be interpreted with caution given the cross-sectional design and limited number of events. These findings support the potential role of retinal vascular changes as cross-sectional correlates of systemic vascular health. Longitudinal studies are needed to clarify temporal relationships and causality.


  • Effect of intraocular pressure-lowering agents in highly myopic eyes

    by Jae Hyup Lee, Kunho Bae, Eun Kyoung Lee, Chang Ki Yoon, Kyu Hyung Park, Un Chul Park

    Background/Aims: To evaluate the effects of intraocular pressure (IOP)-lowering agents in highly myopic eyes. Methods: Highly myopic patients with axial length (AL) data for longer than 5 years were included. The changes in IOP, AL and myopic macular complications during the study period were compared between the eyes treated with IOP-lowering agents (treatment group) and those never treated (control group). Results: The treatment (30 eyes of 30 patients) and control (90 eyes of 90 patients) groups had comparable baseline characteristics except for IOP, which was significantly higher in the treatment group (p=0.020). During the study period, the treatment group showed a greater mean percentage decrease in IOP per year compared with the control group (−2.13±2.51% /year vs+0.30 ± 2.31% /year; p<0.001). The mean change in AL per year was +0.032 ± 0.050 mm/year in the treatment group and +0.058 ± 0.054 mm/year in the control group (p=0.022), and the difference was more pronounced in eyes with posterior staphyloma or without dome-shaped macula. The progression of myopic macular degeneration (MMD), myopic retinoschisis and myopic choroidal neovascularisation during the study period was comparable between groups, but the increase in MMD category was significantly less frequent in the treatment group (3.3% vs 20.0%; p=0.040). In the multivariate analysis, use of IOP-lowering agents was associated with both decreased axial elongation and lower likelihood of MMD category increase. Conclusion: IOP-lowering agents may be effective in alleviating axial elongation and progression of myopic maculopathy in highly myopic eyes.

  • Open Access

    Risk of Retinal Vascular Occlusive Disease According to Type and Low-Density Lipoprotein-Cholesterol Control after Acute Coronary Syndrome

    by Minjeong Kim, Ji Hyun Lee, Yun-Hyeong Cho, Kunho Bae, Ju-Yeun Lee

    Kim M, et al. Yonsei Med J. 2025 Dec;66:e115. https://doi.org/10.3349/ymj.2025.0128


  • Open Access

    Changes in retinal vascular geometry after vitrectomy in idiopathic and uveitic epiretinal membrane

    by Chang Hwan Lee, Seung Woo Choi, Eun Kyoung Lee, Chang Ki Yoon, Un Chul Park, Kyu Hyung Park, Sojung Go, Sang Jun Park, Kunho Bae

    Epiretinal membrane (ERM) can arise idiopathically or secondary to uveitis, each driven by distinct pathophysiological mechanisms. Because ERMs exert tractional forces that distort both the retina and its vasculature, vascular geometry can serve as an indirect marker of the tractional forces acting on the retina. This study aimed to evaluate the associations between postoperative changes in retinal vascular geometry and functional outcomes in patients with idiopathic and uveitic epiretinal membrane (iERM and uERM). A retrospective analysis was performed on 72 eyes (42 iERM, 30 uERM) that underwent pars plana vitrectomy for ERM. Retinal vascular geometric parameters, including arterial and venous angles, branching angles, and tortuosity, were quantitatively measured using conventional retinal fundus photographs. Best-corrected visual acuity (BCVA) and central foveal thickness (CFT) were measured preoperatively and at 1 and 6 months postoperatively. The primary outcome was the comparison of postoperative changes in retinal vascular geometry between iERM and uERM, while the secondary outcome assessed improvements in BCVA, M-score, and CFT, and their correlation with vascular changes. Greater postoperative changes in arterial angle were significantly associated with improvements in BCVA and CFT. The iERM group exhibited more pronounced changes in retinal vascular geometry than the uERM group, particularly in branching venous angles and venous tortuosity. These differences became more evident over time. In conclusion, this study demonstrates that postoperative changes in retinal vascular geometry differ between patients with iERM and uERM, with more significant changes observed in iERM. These findings provide valuable insight into the distinct recovery mechanisms in iERM and uERM, emphasizing the roles of tractional forces and inflammation in shaping postoperative outcomes.