Pub Date : 2025-11-17DOI: 10.1016/j.xops.2025.101011
Taiga Inooka MD, PhD , Yuki Kimura MD , Shota Fujikawa , Sayuri Yasuda MD, PhD , Taro Kominami MD, PhD , Tetsuhito Kojima MD, PhD , Shinji Ueno MD, PhD , Yasuki Ito MD, PhD , Koji M. Nishiguchi MD, PhD , Kenya Yuki MD, PhD
Purpose
Adult axial length (AL) elongation in adults is associated with pathologic outcomes; however, population-based longitudinal evidence remains limited, and pragmatic risk profiling is unclear. We aimed to quantify the prevalence and annual rate of AL elongation in adults and identify independent determinants in a population-based health-check cohort.
Design
Retrospective, single-center cohort study.
Subjects
A total of 4016 adults aged 22.4 to 93.0 years (8032 eyes; 21 421 visits) undergoing a Japanese health-check program, with a median follow-up of 5.31 years.
Methods
For each eye, the annual AL change (mm/year) was estimated as the within-eye linear-regression slope and classified as severe (≥0.10), moderate (≥0.05 to <0.10), mild (≥0.00 to <0.05), or nil (<0). Associations were evaluated using a class-weighted proportional-odds ordinal logistic model. Pair-level change-point analysis modeled which eye elongated faster.
Main Outcome Measures
Proportion of eyes by annual AL elongation severity, adjusted odds ratios (ORs) for determinants of more-severe elongation, and the baseline-AL intersection at which the longer eye becomes more likely to elongate faster.
Results
Nil and mild accounted for 98.6% of eyes; moderate and severe were uncommon (1.3% and 0.2%). Independent determinants of greater severity included longer baseline AL (OR, 1.34 per 1 mm), larger interocular difference in AL (OR, 7.18 per 1 mm), myopic maculopathy including tessellated fundus (OR, 2.79), and female sex (OR, 3.05). Change-point analysis identified an intersection near 28.56 mm in the longer eye: below this value, the estimated probability that it would elongate faster was approximately 0.50 (no consistent lateral preference), whereas above it the probability exceeded 0.50 with wide uncertainty at high baseline AL values.
Conclusions
Adult AL elongation is uncommon and slow; risk is concentrated in eyes with longer AL, greater axial asymmetry, myopic maculopathy, and in female adults. These readily measured features can inform follow-up decisions in health-check settings; pair-level estimates around 28.6 mm may help prioritize eyes for follow-up but should be interpreted cautiously.
Financial Disclosure(s)
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Risk Factors for Adult Axial Length Elongation: A 5-Year Population-Based Cohort Study","authors":"Taiga Inooka MD, PhD , Yuki Kimura MD , Shota Fujikawa , Sayuri Yasuda MD, PhD , Taro Kominami MD, PhD , Tetsuhito Kojima MD, PhD , Shinji Ueno MD, PhD , Yasuki Ito MD, PhD , Koji M. Nishiguchi MD, PhD , Kenya Yuki MD, PhD","doi":"10.1016/j.xops.2025.101011","DOIUrl":"10.1016/j.xops.2025.101011","url":null,"abstract":"<div><h3>Purpose</h3><div>Adult axial length (AL) elongation in adults is associated with pathologic outcomes; however, population-based longitudinal evidence remains limited, and pragmatic risk profiling is unclear. We aimed to quantify the prevalence and annual rate of AL elongation in adults and identify independent determinants in a population-based health-check cohort.</div></div><div><h3>Design</h3><div>Retrospective, single-center cohort study.</div></div><div><h3>Subjects</h3><div>A total of 4016 adults aged 22.4 to 93.0 years (8032 eyes; 21 421 visits) undergoing a Japanese health-check program, with a median follow-up of 5.31 years.</div></div><div><h3>Methods</h3><div>For each eye, the annual AL change (mm/year) was estimated as the within-eye linear-regression slope and classified as severe (≥0.10), moderate (≥0.05 to <0.10), mild (≥0.00 to <0.05), or nil (<0). Associations were evaluated using a class-weighted proportional-odds ordinal logistic model. Pair-level change-point analysis modeled which eye elongated faster.</div></div><div><h3>Main Outcome Measures</h3><div>Proportion of eyes by annual AL elongation severity, adjusted odds ratios (ORs) for determinants of more-severe elongation, and the baseline-AL intersection at which the longer eye becomes more likely to elongate faster.</div></div><div><h3>Results</h3><div>Nil and mild accounted for 98.6% of eyes; moderate and severe were uncommon (1.3% and 0.2%). Independent determinants of greater severity included longer baseline AL (OR, 1.34 per 1 mm), larger interocular difference in AL (OR, 7.18 per 1 mm), myopic maculopathy including tessellated fundus (OR, 2.79), and female sex (OR, 3.05). Change-point analysis identified an intersection near 28.56 mm in the longer eye: below this value, the estimated probability that it would elongate faster was approximately 0.50 (no consistent lateral preference), whereas above it the probability exceeded 0.50 with wide uncertainty at high baseline AL values.</div></div><div><h3>Conclusions</h3><div>Adult AL elongation is uncommon and slow; risk is concentrated in eyes with longer AL, greater axial asymmetry, myopic maculopathy, and in female adults. These readily measured features can inform follow-up decisions in health-check settings; pair-level estimates around 28.6 mm may help prioritize eyes for follow-up but should be interpreted cautiously.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101011"},"PeriodicalIF":4.6,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145798068","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-14DOI: 10.1016/j.xops.2025.101009
Lawrence J. Singerman MD , Jean-Pierre Hubschman MD , Martin Friedlander MD, PhD , Emily Y. Chew MD , Catherine Egan PhD , Muna Bitar PharmD , Thomas M. Aaberg Jr. MD
Objective
To primarily assess long-term safety and retinal imaging outcomes of NT-501 (revakinagene taroretcel-lwey), which releases ciliary neurotrophic factor into the vitreous over an extended time, for treating macular telangiectasia type 2 (MacTel).
Design
Phase I, nonrandomized, multicenter, open-label extension study.
Participants
Six participants with bilateral MacTel who completed the parent 60-month phase I study.
Methods
In the parent study, participants had NT-501 surgically implanted in the study eye. The eye with more advanced disease was determined to be the study eye. For the purposes of this extension study, the fellow eye provided untreated natural history data. The extension study included visits 72, 84, 96, and 108 months postimplantation.
Main Outcome Measures
Safety outcomes included adverse events (AEs), change from baseline in best-corrected visual acuity (BCVA), and the proportions of eyes with ≥10- or ≥15-letter loss in BCVA from baseline. Retinal imaging variables included change from baseline in ellipsoid zone (EZ) (inner segment/outer segment) area loss and proportion of study eyes with ≥35% increase from baseline in EZ area loss.
Results
All implants were retained through the final study visit. All ocular treatment-emergent AEs were mild to moderate; none resulted in study discontinuation. No study eyes had ≥15-letter loss in BCVA from baseline at any study visit. Similarly, no study eyes had ≥10-letter loss at months 72, 84, and 96; 1 study eye (17%) experienced it at month 108. The portion of study eyes with a ≥35% increase in EZ area loss from baseline was lower (range, 50%–60%) relative to fellow eyes (range, 75%–100%).
Conclusions
Over the 9-year follow-up period, NT-501 was well tolerated and safe. Further studies are ongoing to investigate the long-term efficacy of NT-501 for treating MacTel.
Financial Disclosure(s)
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Results from a Phase I Extension Study of Ciliary Neurotrophic Factor in Patients with Macular Telangiectasia Type 2","authors":"Lawrence J. Singerman MD , Jean-Pierre Hubschman MD , Martin Friedlander MD, PhD , Emily Y. Chew MD , Catherine Egan PhD , Muna Bitar PharmD , Thomas M. Aaberg Jr. MD","doi":"10.1016/j.xops.2025.101009","DOIUrl":"10.1016/j.xops.2025.101009","url":null,"abstract":"<div><h3>Objective</h3><div>To primarily assess long-term safety and retinal imaging outcomes of NT-501 (revakinagene taroretcel-lwey), which releases ciliary neurotrophic factor into the vitreous over an extended time, for treating macular telangiectasia type 2 (MacTel).</div></div><div><h3>Design</h3><div>Phase I, nonrandomized, multicenter, open-label extension study.</div></div><div><h3>Participants</h3><div>Six participants with bilateral MacTel who completed the parent 60-month phase I study.</div></div><div><h3>Methods</h3><div>In the parent study, participants had NT-501 surgically implanted in the study eye. The eye with more advanced disease was determined to be the study eye. For the purposes of this extension study, the fellow eye provided untreated natural history data. The extension study included visits 72, 84, 96, and 108 months postimplantation.</div></div><div><h3>Main Outcome Measures</h3><div>Safety outcomes included adverse events (AEs), change from baseline in best-corrected visual acuity (BCVA), and the proportions of eyes with ≥10- or ≥15-letter loss in BCVA from baseline. Retinal imaging variables included change from baseline in ellipsoid zone (EZ) (inner segment/outer segment) area loss and proportion of study eyes with ≥35% increase from baseline in EZ area loss.</div></div><div><h3>Results</h3><div>All implants were retained through the final study visit. All ocular treatment-emergent AEs were mild to moderate; none resulted in study discontinuation. No study eyes had ≥15-letter loss in BCVA from baseline at any study visit. Similarly, no study eyes had ≥10-letter loss at months 72, 84, and 96; 1 study eye (17%) experienced it at month 108. The portion of study eyes with a ≥35% increase in EZ area loss from baseline was lower (range, 50%–60%) relative to fellow eyes (range, 75%–100%).</div></div><div><h3>Conclusions</h3><div>Over the 9-year follow-up period, NT-501 was well tolerated and safe. Further studies are ongoing to investigate the long-term efficacy of NT-501 for treating MacTel.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101009"},"PeriodicalIF":4.6,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145798001","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-12DOI: 10.1016/j.xops.2025.100999
Jost B. Jonas MD , Rahul A. Jonas MD , Mukharram M. Bikbov MD, PhD , Gyulli M. Kazakbaeva MD , Ellina M. Iakupova MD , Ya Xing Wang MD , Vinay Nangia MD , Songhomitra Panda-Jonas MD
Objective
To determine the cutoff value in axial length between moderate myopia versus high myopia in dependence on the prevalence of myopic macular degeneration (MMD).
Design
Population-based studies conducted in Russia, China, and India.
Participants
The project included the population-based investigations of the Beijing Eye Study (n = 3325; age: 40+ years), Russian Ural Eye and Medical Study (n = 5586 participants; age: 40+ years), Ural Very Old Study (n = 541; age: 85+ years) and Ural Children Eye Study (n = 4255; age: 6+ years), and Central India Eye and Medical Study (n = 4467; age: 30+ years).
Methods
The participants underwent a series of general medical and ophthalmic examinations, including fundus photography and ocular biometry. Myopic macular degeneration was defined according to the Meta-analysis for Pathologic Myopia Study Group.
Main Outcome Measures
Prevalence of MMD in dependence on axial length.
Results
The total study population included 36 123 eyes (18 471 individuals) (age: 47.4 ± 23.4 years; range: 6–100 years) (axial length: 23.2 ± 1.1 mm; range: 18.22–34.20 mm). In the total study population, higher MMD stage was associated with longer axial length (β: 0.55; B: 0.14; 95% confidence interval [CI]: 0.14–0.15; P < 0.001), older age (β: 0.09; B: 0.001; 95% CI: 0.001–0.001; P < 0.001), female sex (β: 0.12; B: 0.07; 95% CI: 0.06–0.07; P < 0.001), and Indian ethnicity (β: 0.10; B: 0.07; 95% CI: 0.06–0.07; P < 0.001). Higher prevalence of MMD stage 2+ and 3+ correlated with longer axial length (odds ratio [OR]: 9.10; 95% CI: 6.93–11.9; P < 0.001 and OR: 6.90; 95% CI: 5.14–9.26; P < 0.001, respectively), older age (OR: 1.06; 95% CI: 1.04–1.08; P < 0.001 and OR: 1.08; 95% CI: 1.05–1.11; P < 0.001, respectively), and Indian ethnicity (OR: 6.96; 95% CI: 2.70–17.9; P < 0.001 and OR: 3.69; 95% CI: 1.26–10.8; P = 0.02, respectively). The turning points of the regression curves of the associations of axial length with prevalence of MMD stage 2+ and 3+ were located at axial length values of between 26.0 and 26.5 mm, respectively.
Conclusions
The axial length-related cutoff for moderate versus high myopia was located approximately at 26.0 and 26.5 mm for the prevalence of MMD stage 2+ and 3+, respectively, with higher values for younger individuals. Myopic macular degeneration prevalence was higher in the Indian cohort.
Financial Disclosure(s)
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Differentiation Between Moderate Versus High Myopia: The 2-Continent Eye Study","authors":"Jost B. Jonas MD , Rahul A. Jonas MD , Mukharram M. Bikbov MD, PhD , Gyulli M. Kazakbaeva MD , Ellina M. Iakupova MD , Ya Xing Wang MD , Vinay Nangia MD , Songhomitra Panda-Jonas MD","doi":"10.1016/j.xops.2025.100999","DOIUrl":"10.1016/j.xops.2025.100999","url":null,"abstract":"<div><h3>Objective</h3><div>To determine the cutoff value in axial length between moderate myopia versus high myopia in dependence on the prevalence of myopic macular degeneration (MMD).</div></div><div><h3>Design</h3><div>Population-based studies conducted in Russia, China, and India.</div></div><div><h3>Participants</h3><div>The project included the population-based investigations of the Beijing Eye Study (n = 3325; age: 40+ years), Russian Ural Eye and Medical Study (n = 5586 participants; age: 40+ years), Ural Very Old Study (n = 541; age: 85+ years) and Ural Children Eye Study (n = 4255; age: 6+ years), and Central India Eye and Medical Study (n = 4467; age: 30+ years).</div></div><div><h3>Methods</h3><div>The participants underwent a series of general medical and ophthalmic examinations, including fundus photography and ocular biometry. Myopic macular degeneration was defined according to the Meta-analysis for Pathologic Myopia Study Group.</div></div><div><h3>Main Outcome Measures</h3><div>Prevalence of MMD in dependence on axial length.</div></div><div><h3>Results</h3><div>The total study population included 36 123 eyes (18 471 individuals) (age: 47.4 ± 23.4 years; range: 6–100 years) (axial length: 23.2 ± 1.1 mm; range: 18.22–34.20 mm). In the total study population, higher MMD stage was associated with longer axial length (β: 0.55; B: 0.14; 95% confidence interval [CI]: 0.14–0.15; <em>P</em> < 0.001), older age (β: 0.09; B: 0.001; 95% CI: 0.001–0.001; <em>P</em> < 0.001), female sex (β: 0.12; B: 0.07; 95% CI: 0.06–0.07; <em>P</em> < 0.001), and Indian ethnicity (β: 0.10; B: 0.07; 95% CI: 0.06–0.07; <em>P</em> < 0.001). Higher prevalence of MMD stage 2+ and 3+ correlated with longer axial length (odds ratio [OR]: 9.10; 95% CI: 6.93–11.9; <em>P</em> < 0.001 and OR: 6.90; 95% CI: 5.14–9.26; <em>P</em> < 0.001, respectively), older age (OR: 1.06; 95% CI: 1.04–1.08; <em>P</em> < 0.001 and OR: 1.08; 95% CI: 1.05–1.11; <em>P</em> < 0.001, respectively), and Indian ethnicity (OR: 6.96; 95% CI: 2.70–17.9; <em>P</em> < 0.001 and OR: 3.69; 95% CI: 1.26–10.8; <em>P</em> = 0.02, respectively). The turning points of the regression curves of the associations of axial length with prevalence of MMD stage 2+ and 3+ were located at axial length values of between 26.0 and 26.5 mm, respectively.</div></div><div><h3>Conclusions</h3><div>The axial length-related cutoff for moderate versus high myopia was located approximately at 26.0 and 26.5 mm for the prevalence of MMD stage 2+ and 3+, respectively, with higher values for younger individuals. Myopic macular degeneration prevalence was higher in the Indian cohort.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 100999"},"PeriodicalIF":4.6,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145798069","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-12DOI: 10.1016/j.xops.2025.101008
Kyle Bolo MD , Tran Huy Nguyen MS , Sreenidhi Iyengar MS , Zhiwei Li MS , Van Nguyen MD , Brandon J. Wong MD , Jiun L. Do MD, PhD , Jose-Luis Ambite PhD , Carl Kesselman PhD , Lauren P. Daskivich MD , Benjamin Y. Xu MD, PhD
Purpose
To compare the performance of a vision transformer-based foundation model (RETFound) and a supervised convolutional neural network (VGG-19) for detecting referable glaucoma from fundus photographs.
Design
An evaluation of diagnostic technology.
Participants
Six thousand one hundred sixteen participants from the Los Angeles County Department of Health Services Teleretinal Screening Program.
Methods
Fundus photographs were labeled for referable glaucoma (cup-to-disc ratio ≥0.6) by certified optometrists. Four deep learning models were trained on cropped and uncropped images (training N = 8996; validation N = 3002) using 2 architectures: RETFound, a vision transformer with self-supervised pretraining on fundus photographs, and VGG-19. Models were evaluated on a held-out test set (N = 1000) labeled by glaucoma specialists and an external test set (N = 300) from University of Southern California clinics. Performance was assessed while varying training set size and stratifying by demographic factors. xRAI was used for saliency mapping.
Main Outcome Measures
Area under the receiver operating characteristic curve (AUC–ROC) and threshold-specific metrics.
Results
The cropped image VGG-19 model achieved the highest AUC–ROC (0.924 [0.907–0.940]), which was comparable (P = 0.07) to the cropped image RETFound model (0.911 [0.892–0.930]), which achieved the highest Youden-optimal performance (sensitivity 82.6% and specificity 88.2%) and F1 score (0.801). Cropped image models outperformed their uncropped counterparts (RETFound 0.889 [0.868–0.909], VGG-19 0.898 [0.879–0.917]) within each architecture (P < 0.001 for AUC–ROC comparisons). The uncropped image RETFound model performed best on external data (0.886 [0.849–0.924] vs. the next-highest 0.797 [0.746–0.848], P < 0.001 for AUC–ROC comparisons). RETFound models had a performance advantage when trained on smaller datasets (N < 2000 images), and the cropped image RETFound model performed consistently across ethnic groups (P = 0.20), whereas the others did not (P < 0.04). Performance did not vary by age or gender. Saliency maps for both architectures consistently included the optic nerve.
Conclusions
Although both RETFound and VGG-19 models performed well for classification of referable glaucoma, foundation models may be preferable when training data are limited and when domain shift is expected. Training models using images cropped to the region of the optic nerve improves performance regardless of architecture but may reduce model generalizability.
Financial Disclosure(s)
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Comparison of RETFound and a Supervised Convolutional Neural Network for Detection of Referable Glaucoma from Fundus Photographs","authors":"Kyle Bolo MD , Tran Huy Nguyen MS , Sreenidhi Iyengar MS , Zhiwei Li MS , Van Nguyen MD , Brandon J. Wong MD , Jiun L. Do MD, PhD , Jose-Luis Ambite PhD , Carl Kesselman PhD , Lauren P. Daskivich MD , Benjamin Y. Xu MD, PhD","doi":"10.1016/j.xops.2025.101008","DOIUrl":"10.1016/j.xops.2025.101008","url":null,"abstract":"<div><h3>Purpose</h3><div>To compare the performance of a vision transformer-based foundation model (RETFound) and a supervised convolutional neural network (VGG-19) for detecting referable glaucoma from fundus photographs.</div></div><div><h3>Design</h3><div>An evaluation of diagnostic technology.</div></div><div><h3>Participants</h3><div>Six thousand one hundred sixteen participants from the Los Angeles County Department of Health Services Teleretinal Screening Program.</div></div><div><h3>Methods</h3><div>Fundus photographs were labeled for referable glaucoma (cup-to-disc ratio ≥0.6) by certified optometrists. Four deep learning models were trained on cropped and uncropped images (training N = 8996; validation N = 3002) using 2 architectures: RETFound, a vision transformer with self-supervised pretraining on fundus photographs, and VGG-19. Models were evaluated on a held-out test set (N = 1000) labeled by glaucoma specialists and an external test set (N = 300) from University of Southern California clinics. Performance was assessed while varying training set size and stratifying by demographic factors. xRAI was used for saliency mapping.</div></div><div><h3>Main Outcome Measures</h3><div>Area under the receiver operating characteristic curve (AUC–ROC) and threshold-specific metrics.</div></div><div><h3>Results</h3><div>The cropped image VGG-19 model achieved the highest AUC–ROC (0.924 [0.907–0.940]), which was comparable (<em>P</em> = 0.07) to the cropped image RETFound model (0.911 [0.892–0.930]), which achieved the highest Youden-optimal performance (sensitivity 82.6% and specificity 88.2%) and F1 score (0.801). Cropped image models outperformed their uncropped counterparts (RETFound 0.889 [0.868–0.909], VGG-19 0.898 [0.879–0.917]) within each architecture (<em>P</em> < 0.001 for AUC–ROC comparisons). The uncropped image RETFound model performed best on external data (0.886 [0.849–0.924] vs. the next-highest 0.797 [0.746–0.848], <em>P</em> < 0.001 for AUC–ROC comparisons). RETFound models had a performance advantage when trained on smaller datasets (N < 2000 images), and the cropped image RETFound model performed consistently across ethnic groups (<em>P</em> = 0.20), whereas the others did not (<em>P</em> < 0.04). Performance did not vary by age or gender. Saliency maps for both architectures consistently included the optic nerve.</div></div><div><h3>Conclusions</h3><div>Although both RETFound and VGG-19 models performed well for classification of referable glaucoma, foundation models may be preferable when training data are limited and when domain shift is expected. Training models using images cropped to the region of the optic nerve improves performance regardless of architecture but may reduce model generalizability.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101008"},"PeriodicalIF":4.6,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.xops.2025.101004
Taiga Inooka MD, PhD , Hikaru Ota MD, PhD , Yosuke Taki MD, PhD, Sayuri Yasuda MD, PhD, Ai Fujita Sajiki MD, PhD, Ayana Suzumura MD, PhD, Hideyuki Shimizu MD, PhD, Jun Takeuchi MD, PhD, Ryo Tomita MD, PhD, Taro Kominami MD, PhD, Hiroaki Ushida MD, PhD, Kenya Yuki MD, PhD, Koji M. Nishiguchi MD, PhD
<div><h3>Objective</h3><div>Artificial intelligence–powered large language models (LLMs) are increasingly applied in health care. However, studies in ophthalmology assessing whether LLMs can improve the accuracy of complex differential diagnoses in clinical cases, or which levels of clinical experience benefit most from their use, remain lacking. This study assessed the effectiveness of ChatGPT-4o, an LLM-driven chatbot, in enhancing ophthalmologists' clinical reasoning using original scenarios.</div></div><div><h3>Design</h3><div>Prospective study.</div></div><div><h3>Subjects</h3><div>Ten original ophthalmic clinical scenarios with open-ended questions were developed, covering the following subspecialties: oculoplastic and orbital disease, glaucoma, inherited retinal disease, macular disease, neuro-ophthalmology, ocular surface, pediatric ophthalmology, retinal vascular disease, strabismus, and uveitis.</div></div><div><h3>Methods</h3><div>Responses to each clinical scenario were collected from 20 ophthalmologists (10 residents and 10 board-certified ophthalmologists) and ChatGPT-4o. Ophthalmologists subsequently revised their answers with assistance from ChatGPT-4o. All responses were anonymized and independently evaluated by 3 attending ophthalmologists based on 4 metrics: coherency, factuality, comprehensiveness, and safety (each on a 5-point scale).</div></div><div><h3>Main Outcome Measures</h3><div>The median total scores for each group in coherency, factuality, comprehensiveness, and safety (maximum of 15 points each).</div></div><div><h3>Results</h3><div>Assistance from ChatGPT-4o significantly improved evaluation scores for coherency, comprehensiveness, and safety among both residents and board-certified ophthalmologists (all, <em>P</em> < 0.001). However, factuality scores showed no significant improvements (<em>P</em> = 0.114 and 0.839, respectively). Although ChatGPT-4o assistance increased citation frequency (residents: 0.24–0.98 per response, board-certified ophthalmologists: 0.12–0.68 per response, both <em>P</em> < 0.05), approximately 44% of these additional citations were identified as hallucinated references, nonexistent, or incorrect citations. Notably, ChatGPT-4o assistance led to a significant increase in variability for factuality and safety scores in both groups (Brown–Forsythe test, all <em>P</em> < 0.05), whereas it decreased variability for coherency and comprehensiveness, with the reduction statistically significant among residents (<em>P</em> = 0.008 and <em>P</em> = 0.006, respectively).</div></div><div><h3>Conclusions</h3><div>ChatGPT-4o effectively enhanced diagnostic reasoning and response quality, particularly among ophthalmology residents. However, successful integration into clinical education and practice requires careful management of increased variability in factuality and safety. This issue could be addressed by implementing strategies such as advanced retrieval-augmented generation systems to ens
{"title":"Evolving Consultation: Enhancing Ophthalmic Diagnostic Performance Using Large Language Model","authors":"Taiga Inooka MD, PhD , Hikaru Ota MD, PhD , Yosuke Taki MD, PhD, Sayuri Yasuda MD, PhD, Ai Fujita Sajiki MD, PhD, Ayana Suzumura MD, PhD, Hideyuki Shimizu MD, PhD, Jun Takeuchi MD, PhD, Ryo Tomita MD, PhD, Taro Kominami MD, PhD, Hiroaki Ushida MD, PhD, Kenya Yuki MD, PhD, Koji M. Nishiguchi MD, PhD","doi":"10.1016/j.xops.2025.101004","DOIUrl":"10.1016/j.xops.2025.101004","url":null,"abstract":"<div><h3>Objective</h3><div>Artificial intelligence–powered large language models (LLMs) are increasingly applied in health care. However, studies in ophthalmology assessing whether LLMs can improve the accuracy of complex differential diagnoses in clinical cases, or which levels of clinical experience benefit most from their use, remain lacking. This study assessed the effectiveness of ChatGPT-4o, an LLM-driven chatbot, in enhancing ophthalmologists' clinical reasoning using original scenarios.</div></div><div><h3>Design</h3><div>Prospective study.</div></div><div><h3>Subjects</h3><div>Ten original ophthalmic clinical scenarios with open-ended questions were developed, covering the following subspecialties: oculoplastic and orbital disease, glaucoma, inherited retinal disease, macular disease, neuro-ophthalmology, ocular surface, pediatric ophthalmology, retinal vascular disease, strabismus, and uveitis.</div></div><div><h3>Methods</h3><div>Responses to each clinical scenario were collected from 20 ophthalmologists (10 residents and 10 board-certified ophthalmologists) and ChatGPT-4o. Ophthalmologists subsequently revised their answers with assistance from ChatGPT-4o. All responses were anonymized and independently evaluated by 3 attending ophthalmologists based on 4 metrics: coherency, factuality, comprehensiveness, and safety (each on a 5-point scale).</div></div><div><h3>Main Outcome Measures</h3><div>The median total scores for each group in coherency, factuality, comprehensiveness, and safety (maximum of 15 points each).</div></div><div><h3>Results</h3><div>Assistance from ChatGPT-4o significantly improved evaluation scores for coherency, comprehensiveness, and safety among both residents and board-certified ophthalmologists (all, <em>P</em> < 0.001). However, factuality scores showed no significant improvements (<em>P</em> = 0.114 and 0.839, respectively). Although ChatGPT-4o assistance increased citation frequency (residents: 0.24–0.98 per response, board-certified ophthalmologists: 0.12–0.68 per response, both <em>P</em> < 0.05), approximately 44% of these additional citations were identified as hallucinated references, nonexistent, or incorrect citations. Notably, ChatGPT-4o assistance led to a significant increase in variability for factuality and safety scores in both groups (Brown–Forsythe test, all <em>P</em> < 0.05), whereas it decreased variability for coherency and comprehensiveness, with the reduction statistically significant among residents (<em>P</em> = 0.008 and <em>P</em> = 0.006, respectively).</div></div><div><h3>Conclusions</h3><div>ChatGPT-4o effectively enhanced diagnostic reasoning and response quality, particularly among ophthalmology residents. However, successful integration into clinical education and practice requires careful management of increased variability in factuality and safety. This issue could be addressed by implementing strategies such as advanced retrieval-augmented generation systems to ens","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101004"},"PeriodicalIF":4.6,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145798003","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.xops.2025.101003
Stephen J. Kim MD , Sapna S. Gangaputra MD, MPH , Sara Al Hussein Al Awamlh MD , Leena Choi PhD , Elizabeth A. McNeer MS , Jinsong Sheng MD
<div><h3>Objective</h3><div>We analyzed the cross-sectional associations of 24 inflammatory cytokines with diabetic retinopathy (DR) severity.</div></div><div><h3>Design</h3><div>Prospective, clinical trial at a tertiary academic medical center.</div></div><div><h3>Subjects</h3><div>Three hundred twenty-eight eyes of 164 patients with diabetes with varying severity of DR, including none.</div></div><div><h3>Methods</h3><div>All diabetic eyes had aqueous sampling of both eyes, ETDRS visual acuity, and color fundus photographs. Three groups were enrolled according to grading of baseline color fundus photographs: 23 (46 eyes) patients with diabetes with no DR, 118 (236 eyes) patients with diabetes with moderate nonproliferative DR (NPDR), and 23 (46 eyes) patients with diabetes with proliferative DR. The moderate NPDR group was further subdivided into mild–moderate and moderate–severe groups based on the ETDRS severity scale. Blood was drawn to measure hemoglobin A1c. A microparticle bead-based multiplex assay was used to measure: fibroblast growth factor-2, eotaxin, granulocyte colony-stimulating factor, FMS-like tyrosine kinase 3, GRO, interleukin (IL)-10, monocyte chemotactic protein (MCP)-3, macrophage-derived chemokine, soluble CD40L, IL-17A, IL-1 receptor antagonist, IL-1β, IL-2, IL-4, IL-6, IL-8, induced protein 10, MCP-1, macrophage inflammatory protein-1β, tumor necrosis factor-α, VEGF-A, regulated on activation normal T expressed and secreted, and platelet-derived growth factor-AA and -AB/BB. Triplicate testing of all cytokines was performed.</div></div><div><h3>Main Outcome Measures</h3><div>Aqueous cytokines, DR severity, hemoglobin A1c.</div></div><div><h3>Results</h3><div>Median and interquartile ranges of VEGF-A by grade of eye were 100.57 (80.93–145.67), 153.40 (112.86–206.24), 223.45 (135.27–319.21), and 295.60 (177.46–388.89) pg/mL among no DR, mild–moderate NPDR, moderate–severe NPDR, and proliferative DR groups, respectively. Median and interquartile ranges of IL-6 were 5.45 (3.16–7.86), 8.28 (4.78–20.68), 12.80 (8.24–27.49), and 17.14 (10.31–52.61) pg/mL and of IL-8 were 7.06 (4.10–13.08), 10.53 (6.45–15.85), 16.25 (11.61–24.10), and 20.61 (14.00–27.65) pg/mL among no DR, mild–moderate NPDR, moderate–severe NPDR, and proliferative DR groups, respectively. Similar positive linear relationships were seen with IL-4 and MCP-1. Compared with the no DR group, significant progressive odds ratios were observed among all DR severity groups for VEGF-A, IL-6, IL-8, and IL-4. Significant progressive odds ratios from mild–moderate NPDR to moderate–severe NPDR were observed for FMS-like tyrosine kinase 3, IL-10, induced protein 10, MCP-1, macrophage-derived chemokine, and platelet-derived growth factor-AA. The majority of cytokines demonstrated no relationship with hemoglobin A1c.</div></div><div><h3>Conclusions</h3><div>We report that several key inflammatory cytokines demonstrate cross-sectional associations with DR severity. Our results le
{"title":"INflammatory MediatorS in the PathophysIology of Diabetic REtinopathy Study","authors":"Stephen J. Kim MD , Sapna S. Gangaputra MD, MPH , Sara Al Hussein Al Awamlh MD , Leena Choi PhD , Elizabeth A. McNeer MS , Jinsong Sheng MD","doi":"10.1016/j.xops.2025.101003","DOIUrl":"10.1016/j.xops.2025.101003","url":null,"abstract":"<div><h3>Objective</h3><div>We analyzed the cross-sectional associations of 24 inflammatory cytokines with diabetic retinopathy (DR) severity.</div></div><div><h3>Design</h3><div>Prospective, clinical trial at a tertiary academic medical center.</div></div><div><h3>Subjects</h3><div>Three hundred twenty-eight eyes of 164 patients with diabetes with varying severity of DR, including none.</div></div><div><h3>Methods</h3><div>All diabetic eyes had aqueous sampling of both eyes, ETDRS visual acuity, and color fundus photographs. Three groups were enrolled according to grading of baseline color fundus photographs: 23 (46 eyes) patients with diabetes with no DR, 118 (236 eyes) patients with diabetes with moderate nonproliferative DR (NPDR), and 23 (46 eyes) patients with diabetes with proliferative DR. The moderate NPDR group was further subdivided into mild–moderate and moderate–severe groups based on the ETDRS severity scale. Blood was drawn to measure hemoglobin A1c. A microparticle bead-based multiplex assay was used to measure: fibroblast growth factor-2, eotaxin, granulocyte colony-stimulating factor, FMS-like tyrosine kinase 3, GRO, interleukin (IL)-10, monocyte chemotactic protein (MCP)-3, macrophage-derived chemokine, soluble CD40L, IL-17A, IL-1 receptor antagonist, IL-1β, IL-2, IL-4, IL-6, IL-8, induced protein 10, MCP-1, macrophage inflammatory protein-1β, tumor necrosis factor-α, VEGF-A, regulated on activation normal T expressed and secreted, and platelet-derived growth factor-AA and -AB/BB. Triplicate testing of all cytokines was performed.</div></div><div><h3>Main Outcome Measures</h3><div>Aqueous cytokines, DR severity, hemoglobin A1c.</div></div><div><h3>Results</h3><div>Median and interquartile ranges of VEGF-A by grade of eye were 100.57 (80.93–145.67), 153.40 (112.86–206.24), 223.45 (135.27–319.21), and 295.60 (177.46–388.89) pg/mL among no DR, mild–moderate NPDR, moderate–severe NPDR, and proliferative DR groups, respectively. Median and interquartile ranges of IL-6 were 5.45 (3.16–7.86), 8.28 (4.78–20.68), 12.80 (8.24–27.49), and 17.14 (10.31–52.61) pg/mL and of IL-8 were 7.06 (4.10–13.08), 10.53 (6.45–15.85), 16.25 (11.61–24.10), and 20.61 (14.00–27.65) pg/mL among no DR, mild–moderate NPDR, moderate–severe NPDR, and proliferative DR groups, respectively. Similar positive linear relationships were seen with IL-4 and MCP-1. Compared with the no DR group, significant progressive odds ratios were observed among all DR severity groups for VEGF-A, IL-6, IL-8, and IL-4. Significant progressive odds ratios from mild–moderate NPDR to moderate–severe NPDR were observed for FMS-like tyrosine kinase 3, IL-10, induced protein 10, MCP-1, macrophage-derived chemokine, and platelet-derived growth factor-AA. The majority of cytokines demonstrated no relationship with hemoglobin A1c.</div></div><div><h3>Conclusions</h3><div>We report that several key inflammatory cytokines demonstrate cross-sectional associations with DR severity. Our results le","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101003"},"PeriodicalIF":4.6,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145798067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.xops.2025.100998
Lynn Kandakji PhD , Shafi Balal MBBS , Aleksander Stupnicki iBSc , Siyin Liu MBBS, PhD , Marcello Leucci BOptom , Dan Gore MD , Bruce Allan MD , Nikolas Pontikos PhD
Purpose
To objectively identify subclinical keratoconus (SKC) from a large sample of healthy and keratoconus (KC) patients via a data-driven framework on corneal imaging data from an anterior-segment OCT (AS-OCT) device (MS-39, CSO Italia).
Design
A retrospective cohort study.
Subjects
At 2 sites within the Moorfields Eye Hospital network in London, United Kingdom, 25 816 corneal scans from 5005 patients, including 3605 with KC and 1400 healthy control patients, were acquired between 2020 and 2024.
Methods
Principal component analysis (PCA) followed by Gaussian mixture modeling (GMM) was applied to AS-OCT–derived data, including 20 KC indices and patient age, to identify SKC eyes, which were then statistically compared against healthy and KC eyes. Subclinical KC eyes were also validated against external systems including same-day Pentacam (Oculus Optikgeräte) scans, Belin-Ambrosio’s ABCD system, KC progression criteria determined by a panel of corneal specialists, and the Moorfields Corneal Cross-linking (CXL) Risk Calculator.
Main Outcome Measures
Detection of SKC and progression of these eyes to clinically diagnosable KC over time.
Results
The GMM identified 166 eyes from 161 patients with distinct structural differences between healthy and KC eyes. These eyes clustered in the morphometric transition zone in PCA space and were predominantly classified as ABCD stage 0. However, they demonstrated asymmetry with their fellow eye, higher predicted CXL risk at 1–4 years (P < 0.001), and faster progression to KC (log-rank P < 0.0001) compared with healthy eyes. Among SKC eyes with longitudinal data, 72.7% met Global Consensus criteria for progression.
Conclusions
Subclinical KC remains challenging to detect, and although classic staging such as ABCD retains clinical utility, it is insufficient for early disease detection. Principal component analysis followed by GMM classification on a multidimensional AS-OCT dataset identifies a distinct and high-risk SKC group. This semisupervised framework offers a complementary tool for early risk stratification and can be applied to new patients via projection into the learned PCA space and computation of KC probability. Threshold values corresponding to the 25th and 75th percentiles of KC probability for each parameter may serve as clinical context for flagging eyes when multiple features fall in the atypical range.
Financial Disclosures
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Data-Driven Detection of Subclinical Keratoconus via Semi-Supervised Clustering of Multidimensional Corneal Biomarkers","authors":"Lynn Kandakji PhD , Shafi Balal MBBS , Aleksander Stupnicki iBSc , Siyin Liu MBBS, PhD , Marcello Leucci BOptom , Dan Gore MD , Bruce Allan MD , Nikolas Pontikos PhD","doi":"10.1016/j.xops.2025.100998","DOIUrl":"10.1016/j.xops.2025.100998","url":null,"abstract":"<div><h3>Purpose</h3><div>To objectively identify subclinical keratoconus (SKC) from a large sample of healthy and keratoconus (KC) patients via a data-driven framework on corneal imaging data from an anterior-segment OCT (AS-OCT) device (MS-39, CSO Italia).</div></div><div><h3>Design</h3><div>A retrospective cohort study.</div></div><div><h3>Subjects</h3><div>At 2 sites within the Moorfields Eye Hospital network in London, United Kingdom, 25 816 corneal scans from 5005 patients, including 3605 with KC and 1400 healthy control patients, were acquired between 2020 and 2024.</div></div><div><h3>Methods</h3><div>Principal component analysis (PCA) followed by Gaussian mixture modeling (GMM) was applied to AS-OCT–derived data, including 20 KC indices and patient age, to identify SKC eyes, which were then statistically compared against healthy and KC eyes. Subclinical KC eyes were also validated against external systems including same-day Pentacam (Oculus Optikgeräte) scans, Belin-Ambrosio’s ABCD system, KC progression criteria determined by a panel of corneal specialists, and the Moorfields Corneal Cross-linking (CXL) Risk Calculator.</div></div><div><h3>Main Outcome Measures</h3><div>Detection of SKC and progression of these eyes to clinically diagnosable KC over time.</div></div><div><h3>Results</h3><div>The GMM identified 166 eyes from 161 patients with distinct structural differences between healthy and KC eyes. These eyes clustered in the morphometric transition zone in PCA space and were predominantly classified as ABCD stage 0. However, they demonstrated asymmetry with their fellow eye, higher predicted CXL risk at 1–4 years (<em>P</em> < 0.001), and faster progression to KC (log-rank <em>P</em> < 0.0001) compared with healthy eyes. Among SKC eyes with longitudinal data, 72.7% met Global Consensus criteria for progression.</div></div><div><h3>Conclusions</h3><div>Subclinical KC remains challenging to detect, and although classic staging such as ABCD retains clinical utility, it is insufficient for early disease detection. Principal component analysis followed by GMM classification on a multidimensional AS-OCT dataset identifies a distinct and high-risk SKC group. This semisupervised framework offers a complementary tool for early risk stratification and can be applied to new patients via projection into the learned PCA space and computation of KC probability. Threshold values corresponding to the 25th and 75th percentiles of KC probability for each parameter may serve as clinical context for flagging eyes when multiple features fall in the atypical range.</div></div><div><h3>Financial Disclosures</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 100998"},"PeriodicalIF":4.6,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145749032","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.xops.2025.101005
Yusong Zhou MD , Qingqing Ye MB , Xuan Qiu MD, PhD , Zixuan Xu MD, Yunsi He MD, Ying Yao MD, Yangfei Pang MD, Wentong Yu MD, Yudan Zhong MM, Qiuying Li MM, Lei Feng MB, Junpeng Yuan BS, Yun Wen MD, PhD, Zhonghao Wang MD, Jinrong Li MD, PhD
Purpose
Perceptual learning (PL) is a novel therapeutic approach for the treatment of amblyopia. This study evaluated whether extending the duration of PL could yield sustained therapeutic benefits in amblyopia management.
Design
A retrospective observational design.
Methods
Analysis included 93 of 100 patients who completed 6-month lateral masking PL and had 3- or 6-month follow-up data. Missing data were handled using multiple imputation by chained equations. Longitudinal changes in visual function—including best-corrected visual acuity (BCVA), contrast sensitivity function (CSF), and stereopsis—were evaluated using a linear mixed-effects model. To identify factors associated with improvements in BCVA and area under the log CSF (AULCSF), multivariable linear regression models were constructed, incorporating all relevant covariates selected based on clinical significance and evidence from existing literature. Only 1 amblyopic eye per patient was included in the analysis of BCVA and AULCSF. In cases of bilateral amblyopia, the worse eye was selected for the primary analysis. Sensitivity analyses were conducted in 2 ways: (1) using measurements from the better eye and (2) using datasets in which missing values had been imputed based on logical rules; the missing value was replaced with the worse outcome of the 2 observed values from the remaining time points.
Main Outcome Measures
Measurements included BCVA, CSF, and stereoacuity.
Results
A total of 93 participants (46 males and 47 females) were enrolled, with a mean age of 15.3 ± 8.3 years. Amblyopia subtypes included: isoametropic (n = 8), anisometropic (n = 63), strabismic (n = 12), deprivation (n = 1), and mixed (n = 9). Analysis revealed significant and sustained improvements in both BCVA and the AULCSF. Near stereopsis also improved following 3 months of PL training. However, further extension of the training duration did not result in additional significant gains in stereopsis. Multivariable linear regression analysis indicated that initial baseline visual function and history of occlusion therapy were the primary factors associated with the improvement of BCVA and AULCSF.
Conclusions
Perceptual learning treatment can improve the visual function of amblyopia patients. Considering that extending the duration of PL still resulted in measurable visual improvements, a 6-month training of PL appears to be necessary.
Financial Disclosure(s)
The authors have no proprietary or commercial interest in any materials discussed in this article.
{"title":"Extending Treatment Duration in Perceptual Learning for Amblyopia","authors":"Yusong Zhou MD , Qingqing Ye MB , Xuan Qiu MD, PhD , Zixuan Xu MD, Yunsi He MD, Ying Yao MD, Yangfei Pang MD, Wentong Yu MD, Yudan Zhong MM, Qiuying Li MM, Lei Feng MB, Junpeng Yuan BS, Yun Wen MD, PhD, Zhonghao Wang MD, Jinrong Li MD, PhD","doi":"10.1016/j.xops.2025.101005","DOIUrl":"10.1016/j.xops.2025.101005","url":null,"abstract":"<div><h3>Purpose</h3><div>Perceptual learning (PL) is a novel therapeutic approach for the treatment of amblyopia. This study evaluated whether extending the duration of PL could yield sustained therapeutic benefits in amblyopia management.</div></div><div><h3>Design</h3><div>A retrospective observational design.</div></div><div><h3>Methods</h3><div>Analysis included 93 of 100 patients who completed 6-month lateral masking PL and had 3- or 6-month follow-up data. Missing data were handled using multiple imputation by chained equations. Longitudinal changes in visual function—including best-corrected visual acuity (BCVA), contrast sensitivity function (CSF), and stereopsis—were evaluated using a linear mixed-effects model. To identify factors associated with improvements in BCVA and area under the log CSF (AULCSF), multivariable linear regression models were constructed, incorporating all relevant covariates selected based on clinical significance and evidence from existing literature. Only 1 amblyopic eye per patient was included in the analysis of BCVA and AULCSF. In cases of bilateral amblyopia, the worse eye was selected for the primary analysis. Sensitivity analyses were conducted in 2 ways: (1) using measurements from the better eye and (2) using datasets in which missing values had been imputed based on logical rules; the missing value was replaced with the worse outcome of the 2 observed values from the remaining time points.</div></div><div><h3>Main Outcome Measures</h3><div>Measurements included BCVA, CSF, and stereoacuity.</div></div><div><h3>Results</h3><div>A total of 93 participants (46 males and 47 females) were enrolled, with a mean age of 15.3 ± 8.3 years. Amblyopia subtypes included: isoametropic (n = 8), anisometropic (n = 63), strabismic (n = 12), deprivation (n = 1), and mixed (n = 9). Analysis revealed significant and sustained improvements in both BCVA and the AULCSF. Near stereopsis also improved following 3 months of PL training. However, further extension of the training duration did not result in additional significant gains in stereopsis. Multivariable linear regression analysis indicated that initial baseline visual function and history of occlusion therapy were the primary factors associated with the improvement of BCVA and AULCSF.</div></div><div><h3>Conclusions</h3><div>Perceptual learning treatment can improve the visual function of amblyopia patients. Considering that extending the duration of PL still resulted in measurable visual improvements, a 6-month training of PL appears to be necessary.</div></div><div><h3>Financial Disclosure(s)</h3><div>The authors have no proprietary or commercial interest in any materials discussed in this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101005"},"PeriodicalIF":4.6,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.xops.2025.101006
Sean Chang BS , Alexandra Hong BS , Yevgeniy Sazhnyev MD , Kevin Choy BS , Blythe Durbin-Johnson PhD , Raymond Ko BS , Neha Sarabu BS , Sina Farsiu PhD , Parisa Emami-Naeini MD, MPH , Ala Moshiri MD, PhD , Kareem Moussa MD , Glenn Yiu MD, PhD
Purpose
The suprachoroidal space (SCS) is an emerging route for drug delivery and gene therapy. Here, we analyze demographic and ocular factors associated with SCS visibility and thickness on enhanced-depth imaging OCT (EDI-OCT) in a large cohort of healthy human eyes to better understand variations in macular SCS measurements.
Design
Retrospective, cross-sectional study.
Subjects
Six hundred twenty-four healthy eyes of 624 patients with no retinal pathologies.
Methods
We analyzed EDI-OCT images of eyes with no retinal pathologies from patients seen at the University of California, Davis between September 1, 2014 and December 31, 2023. We performed image segmentation and measured SCS visibility and thickness along the 6-mm segment around the fovea, as well as retinal and choroidal thicknesses. Univariate and multivariate regression analyses were used to determine the association of SCS visibility and thickness with demographic and ocular factors including age, sex, race, refractive error, and retinal and choroidal thicknesses.
Main Outcome Measures
Association of SCS visibility and thickness with demographic and ocular factors.
Results
In healthy subjects (mean age 64.7, range 14–98 years), the choroidal–scleral junction was visible on EDI-OCT in 462 of 624 eyes (74%), among which a hyporeflective SCS layer could be discerned in 214 eyes (46%). The SCS layer was more likely to be present in older (P < 0.001) and White (P = 0.022) patients. In eyes with a detectable SCS layer, median (interquartile range) subfoveal SCS thickness was 34.8 (23.2–46.4) μm, and 32.8 (23.8–41.9) μm across the central macula. Both subfoveal and macular SCS thickness were higher with older age (P < 0.001), while subfoveal SCS thickness was also greater in White patients (P = 0.027).
Conclusions
Suprachoroidal space anatomy varies with age and race. Understanding factors associated with SCS measurements could help inform future research focused on SCS-targeted therapies or patient selection in future clinical trials.
Financial Disclosure(s)
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Demographic and Ocular Factors Associated with Suprachoroidal Space Thickness in Healthy Eyes","authors":"Sean Chang BS , Alexandra Hong BS , Yevgeniy Sazhnyev MD , Kevin Choy BS , Blythe Durbin-Johnson PhD , Raymond Ko BS , Neha Sarabu BS , Sina Farsiu PhD , Parisa Emami-Naeini MD, MPH , Ala Moshiri MD, PhD , Kareem Moussa MD , Glenn Yiu MD, PhD","doi":"10.1016/j.xops.2025.101006","DOIUrl":"10.1016/j.xops.2025.101006","url":null,"abstract":"<div><h3>Purpose</h3><div>The suprachoroidal space (SCS) is an emerging route for drug delivery and gene therapy. Here, we analyze demographic and ocular factors associated with SCS visibility and thickness on enhanced-depth imaging OCT (EDI-OCT) in a large cohort of healthy human eyes to better understand variations in macular SCS measurements.</div></div><div><h3>Design</h3><div>Retrospective, cross-sectional study.</div></div><div><h3>Subjects</h3><div>Six hundred twenty-four healthy eyes of 624 patients with no retinal pathologies.</div></div><div><h3>Methods</h3><div>We analyzed EDI-OCT images of eyes with no retinal pathologies from patients seen at the University of California, Davis between September 1, 2014 and December 31, 2023. We performed image segmentation and measured SCS visibility and thickness along the 6-mm segment around the fovea, as well as retinal and choroidal thicknesses. Univariate and multivariate regression analyses were used to determine the association of SCS visibility and thickness with demographic and ocular factors including age, sex, race, refractive error, and retinal and choroidal thicknesses.</div></div><div><h3>Main Outcome Measures</h3><div>Association of SCS visibility and thickness with demographic and ocular factors.</div></div><div><h3>Results</h3><div>In healthy subjects (mean age 64.7, range 14–98 years), the choroidal–scleral junction was visible on EDI-OCT in 462 of 624 eyes (74%), among which a hyporeflective SCS layer could be discerned in 214 eyes (46%). The SCS layer was more likely to be present in older (<em>P</em> < 0.001) and White (<em>P</em> = 0.022) patients. In eyes with a detectable SCS layer, median (interquartile range) subfoveal SCS thickness was 34.8 (23.2–46.4) μm, and 32.8 (23.8–41.9) μm across the central macula. Both subfoveal and macular SCS thickness were higher with older age (<em>P</em> < 0.001), while subfoveal SCS thickness was also greater in White patients (<em>P</em> = 0.027).</div></div><div><h3>Conclusions</h3><div>Suprachoroidal space anatomy varies with age and race. Understanding factors associated with SCS measurements could help inform future research focused on SCS-targeted therapies or patient selection in future clinical trials.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101006"},"PeriodicalIF":4.6,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145749033","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.xops.2025.101007
Nikoo Hamzeh MD, MPH, Alcina K. Lidder MD, Robert S. Feder MD, Emmanuel A. Sarmiento MD, Rukhsana G. Mirza MD, Avrey J. Thau MD, Angelo P. Tanna MD
Purpose
To assess the quality of Chat Generative Pre-Trained Transformer-4 Omni (ChatGPT-4o) responses to questions submitted by patients through Epic MyChart.
Design
Retrospective cross-sectional study.
Participants
One hundred sixty-five patients who submitted ophthalmology-related questions via Epic MyChart.
Methods
Questions asked by ophthalmology clinic patients related to the subspecialties of glaucoma, retina, and cornea via the Epic MyChart at a single institution were evaluated. Nonclinical questions were excluded. Each question was submitted to ChatGPT-4o twice, first without limitations and then after priming the large language model (LLM) to respond at a sixth-grade reading level. The ChatGPT-4o output and subsequent conversations were graded by 2 independent ophthalmologist reviewers as “accurate and complete,” “incomplete,” or “unacceptable” with respect to the quality of the output. A third subspecialist reviewer provided adjudication in cases of disagreement. Readability of the ChatGPT-4o output was assessed using the Flesch–Kincaid Grade Level and other readability indices.
Main Outcome Measures
Quality and readability of answers generated by ChatGPT-4o.
Results
Two hundred eighty-five queries asked by 165 patients were analyzed. Overall, 220 (77%) responses were graded as accurate and complete, 49 (17%) as incomplete, and 16 (6%) as unacceptable. The initial 2 reviewers agreed in 87% of the responses generated by ChatGPT-4o. The overall mean Flesch–Kincaid reading grade level was 12.1 ± 2.1. When asked to respond at a sixth-grade reading level, 242 (85%) responses were graded as accurate and complete, 38 (13%) were incomplete, and 5 (2%) were graded as unacceptable.
Conclusions
Chat Generative Pre-Trained Transformer-4 Omni usually provides accurate and complete answers to the questions posed by patients to their glaucoma, retina, and cornea subspecialists. A substantial proportion of the responses were, however, graded as incomplete or unacceptable. Chat Generative Pre-Trained Transformer-4 Omni responses required a 12th-grade education level as assessed by Flesch–Kincaid and other readability indices, which may make them difficult for many patients to understand; however, when prompted to do so, the LLM can generate responses at a sixth-grade reading level without a compromise in response quality. Chat Generative Pre-Trained Transformer-4 Omni can potentially be used to answer clinical ophthalmology questions posed by patients; however, additional refinement will be required prior to implementation of such an approach.
Financial Disclosure(s)
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
{"title":"Accuracy and Readability of Chat Generative Pre-Trained Transformer-4 Omni in Answering Ophthalmology Patient Questions","authors":"Nikoo Hamzeh MD, MPH, Alcina K. Lidder MD, Robert S. Feder MD, Emmanuel A. Sarmiento MD, Rukhsana G. Mirza MD, Avrey J. Thau MD, Angelo P. Tanna MD","doi":"10.1016/j.xops.2025.101007","DOIUrl":"10.1016/j.xops.2025.101007","url":null,"abstract":"<div><h3>Purpose</h3><div>To assess the quality of Chat Generative Pre-Trained Transformer-4 Omni (ChatGPT-4o) responses to questions submitted by patients through Epic MyChart.</div></div><div><h3>Design</h3><div>Retrospective cross-sectional study.</div></div><div><h3>Participants</h3><div>One hundred sixty-five patients who submitted ophthalmology-related questions via Epic MyChart.</div></div><div><h3>Methods</h3><div>Questions asked by ophthalmology clinic patients related to the subspecialties of glaucoma, retina, and cornea via the Epic MyChart at a single institution were evaluated. Nonclinical questions were excluded. Each question was submitted to ChatGPT-4o twice, first without limitations and then after priming the large language model (LLM) to respond at a sixth-grade reading level. The ChatGPT-4o output and subsequent conversations were graded by 2 independent ophthalmologist reviewers as “accurate and complete,” “incomplete,” or “unacceptable” with respect to the quality of the output. A third subspecialist reviewer provided adjudication in cases of disagreement. Readability of the ChatGPT-4o output was assessed using the Flesch–Kincaid Grade Level and other readability indices.</div></div><div><h3>Main Outcome Measures</h3><div>Quality and readability of answers generated by ChatGPT-4o.</div></div><div><h3>Results</h3><div>Two hundred eighty-five queries asked by 165 patients were analyzed. Overall, 220 (77%) responses were graded as accurate and complete, 49 (17%) as incomplete, and 16 (6%) as unacceptable. The initial 2 reviewers agreed in 87% of the responses generated by ChatGPT-4o. The overall mean Flesch–Kincaid reading grade level was 12.1 ± 2.1. When asked to respond at a sixth-grade reading level, 242 (85%) responses were graded as accurate and complete, 38 (13%) were incomplete, and 5 (2%) were graded as unacceptable.</div></div><div><h3>Conclusions</h3><div>Chat Generative Pre-Trained Transformer-4 Omni usually provides accurate and complete answers to the questions posed by patients to their glaucoma, retina, and cornea subspecialists. A substantial proportion of the responses were, however, graded as incomplete or unacceptable. Chat Generative Pre-Trained Transformer-4 Omni responses required a 12th-grade education level as assessed by Flesch–Kincaid and other readability indices, which may make them difficult for many patients to understand; however, when prompted to do so, the LLM can generate responses at a sixth-grade reading level without a compromise in response quality. Chat Generative Pre-Trained Transformer-4 Omni can potentially be used to answer clinical ophthalmology questions posed by patients; however, additional refinement will be required prior to implementation of such an approach.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 2","pages":"Article 101007"},"PeriodicalIF":4.6,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145749067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号