Background: Diabetic retinopathy is a leading cause of preventable vision loss, and patients increasingly seek disease related information through online consultations. Large language models may support patient education, but their reliability and usability vary across systems, particularly in disease specific settings. Methods: Thirty common patient questions about diabetic retinopathy were developed from guidelines and organized into five domains: disease overview, screening and diagnosis, treatment and follow up, lifestyle and prevention, and prognosis and complication management. From November 10 to 15, 2025, two researchers independently submitted all questions to five models (ChatGPT-5, DeepSeek-V3.1, Doubao, Wenxinyiyan 4.5 Turbo, and Kimi) on public platforms under identical conditions without system prompts. Chat histories were reset before each question. Response time, response length, structural metrics, and table outputs were extracted. Two retinal specialists rated each answer on a 1 to 5 Likert scale across accuracy, logical consistency, coherence, safety, and content accessibility. Inter rater agreement was assessed with the intraclass correlation coefficient. Group differences were analyzed using analysis of variance or the Kruskal–Wallis H test with Bonferroni corrected pairwise comparisons. Results: Significant between model differences were observed in output efficiency and textual characteristics (all P < 0.001). ChatGPT-5 responded fastest (15.92 ± 4.48 s), whereas Wenxinyiyan 4.5 Turbo and DeepSeek-V3.1 were slowest (41.89 ± 5.09 s and 38.20 ± 2.96 s). DeepSeek-V3.1 generated the longest answers (1396.37 ± 189.23 words), while Kimi produced the shortest (579.40 ± 182.96 words). Only ChatGPT-5 consistently generated structured tables (median 2.00, IQR 1.00-2.00). Content quality differed significantly across all five dimensions (H = 15.34-37.19, all P ≤ 0.004). ChatGPT-5 achieved the highest median scores for accuracy (5.00, IQR 4.00-5.00) and logical consistency (4.50, IQR 4.00-5.00), whereas Kimi showed the lowest accuracy (3.50, IQR 3.00-4.00). The intraclass correlation coefficient indicated good inter rater reliability (0.87). Conclusion: Performance of large language models in diabetic retinopathy patient consultations is model dependent. ChatGPT-5 demonstrated the best overall usability, combining faster responses, clearer structure, and higher factual accuracy. Other Chinese optimized models provided comparable professional information coverage but require improved accessibility and stability for safe patient facing use.

Many major diseases can cause changes in the morphology of blood vessels, and the segmentation of retinal blood vessels is of great significance for preventing these diseases. Obtaining complete, continuous, and high-resolution segmentation results is very challenging due to the diverse structures of retinal tissues, the complex spatial structures of blood vessels, and the presence of many small ships. In recent years, deep learning networks like UNet have been widely used in medical image processing. However, the continuous down-sampling operations in UNet can result in the loss of a significant amount of information. Although skip connections between the encoder and decoder can help address this issue, the encoder features still contain a large amount of irrelevant information that cannot be efficiently utilized by the decoder. To alleviate the irrelevant information, this paper proposes a feature selection module between the decoder and encoder that utilizes the self-attention mechanism of transformers to accurately and efficiently select the relevant encoder features for the decoder. Additionally, a lightweight Residual Global Context module is proposed to obtain dense global contextual information and establish dependencies between pixels, which can effectively preserve vascular details and segment small vessels accurately and continuously. Experimental results on three publicly available color fundus image datasets (DRIVE, CHASE, and STARE) demonstrate that the proposed algorithm outperforms existing methods in terms of both performance metrics and visual quality.

Based on the Dreyfus model of skill acquisition, this article classifies the professional development of ophthalmologists into four stages: novice, advanced beginner, competent, and expert. In this review, artificial intelligence (AI) is operationally defined as data-driven algorithms that enable prediction, perception, and objective assessment from multimodal surgical data. We distinguish AI methods from immersive hardware, such as virtual reality (VR), which serves as a training interface that may or may not incorporate AI-driven assessment and feedback. Accordingly, this manuscript focuses on AI-enabled simulation, computer-vision-based surgical video understanding, and registry/EHR-driven clinical practice and training continuum. At the novice stage, AI-enabled assessment within VR simulation helps trainees form muscle memory and standardized operating habits. This is achieved through haptic-enabled modules and objective performance metrics. For advanced beginners, computer-vision models and attention-visualization techniques support surgical workflow understanding and structured debriefing, assisting trainees in building surgical logic and spatial cognition. When doctors reach the competent stage, AI uses large-scale clinical data to estimate complication risk and support scenario-based crisis training, strengthening complication management and non-technical skills. At the expert stage, AI-assisted surgical video analytics can benchmark technique patterns and surface potential blind spots, facilitating continuous calibration and knowledge sharing. Overall, the evidence to date suggests that AI is best positioned as an assistive tool to augment human learning and decision-making. However, generalizability, interpretability, data governance, and medicolegal accountability remain key barriers to safe and scalable deployment.

Background: Congenital nasolacrimal duct obstruction is a common ocular condition in early infancy and may lead to neonatal dacryocystitis or severe infection if not treated promptly. Short-video platforms such as TikTok are used by young parents to obtain health information, but the quality of related videos remains unclear. This study evaluated the quality of TikTok videos on congenital nasolacrimal duct obstruction and its association with uploader type and user engagement. Methods: We conducted a cross-sectional review of TikTok videos retrieved with predefined keywords. We included 108 videos, classified by uploader type, and extracted characteristics, engagement metrics, and coverage of six content domains. Two attending ophthalmologists independently rated each video using DISCERN, the Global Quality Score, and the Patient Education Materials Assessment Tool for Audiovisual Materials. Group differences and Spearman correlations were analyzed. Results: Among 108 videos, 62 (57.4%) were uploaded by medical professionals and 18 (16.7%) by non-Profit organizations, with a median duration of 48.5 s and median numbers of likes, comments, favorites and shares of 68, 8, 13, and 24, respectively. Videos uploaded by non-Profit organizations and medical professionals achieved substantially higher DISCERN scores (about 59.22 and 50.00 vs. 37.50 and 19.40), Global Quality Scores (median 4.5 and 4.0 vs. 3.0 and 1.0), and Patient Education Materials Assessment Tool for Audiovisual Materials understandability (median 92.31 and 84.62% vs. 73.08 and 53.85%) and actionability (both 75.00% vs. 66.67 and 50.00%) than those uploaded by non-Medical individuals and for-profit organizations (all P < 0.001). Spearman analysis found no significant correlations between overall quality scores and engagement, but duration correlated weakly with some selected quality indicators ( r = 0.17–0.21, P < 0.05). Conclusion: TikTok videos on congenital nasolacrimal duct obstruction show marked heterogeneity in content completeness and educational quality, largely determined by uploader type. Non-Profit organizations and medical professionals produce more reliable, understandable, and actionable videos, but these are not consistently more popular than lower-quality content. Clinicians and institutions should develop guideline-concordant short videos and direct parents toward trustworthy channels, while platforms consider mechanisms to highlight professionally verified pediatric eye-health information.

Background: Integrated traditional Chinese and Western medicine has shown clear advantages in the management of ophthalmic diseases. However, misalignment between traditional Chinese medicine syndrome differentiation and Western pathological classification, overemphasis on Western diagnostic and therapeutic procedures, and insufficient training in traditional Chinese medicine techniques continue to constrain teaching quality in ophthalmology. Methods: This single-center prospective interventional study implemented a three-dimensional integrated traditional Chinese and Western medicine teaching model that combined case-based learning and problem-based learning, and compared teaching outcomes in 156 medical students before and after the intervention. Results: The new teaching model received a mean satisfaction score of 4.54 ± 0.33 on a five-point Likert scale. Compared with baseline, students showed significantly higher classroom participation (92.0 ± 4.5% compared with 64.0 ± 8.5%, p < 0.001), comprehensive examination scores (92.0 ± 4.0 compared with 77.0 ± 5.5, p < 0.001), and overall autonomous learning ability scores (4.54 ± 0.33 compared with 3.20 ± 0.45, p < 0.001). Additional improvements were observed in independent literature review frequency (3.5 ± 0.4 compared with 1.2 ± 0.3 times per week, p < 0.001), acupuncture point location accuracy (95.0% compared with 84.0%, p < 0.001), and Western medicine examination scores (95.5 ± 3.1 compared with 80.5 ± 4.2, p < 0.001). Conclusion: The three-dimensional integrated traditional Chinese and Western medicine teaching model effectively enhanced ophthalmology teaching quality and helped cultivate medical students with integrated traditional Chinese and Western medicine competencies, as reflected by improved student engagement, autonomous learning, and clinical skill mastery.

Background: Digital tools are reshaping public health education and training, yet evidence on whether large language models (LLMs) can generate specialist ophthalmic teaching materials remains limited. High myopia (HM), a vision-threatening condition with long-term management needs and public health relevance, provides a suitable setting for evaluating this capability. This study compared five LLMs in generating HM-related multiple-choice questions (MCQs) for ophthalmic education. Methods: Five LLMs (ChatGPT-5.4, Gemini 3, DeepSeek, Kimi K2.5, and Doubao) completed 60 predefined HM MCQ generation tasks each, yielding 300 MCQs. A standardized blueprint covered four domains: basic knowledge, clinical cases, diagnosis and treatment decision-making, and screening/follow-up management. Objective evaluation included structural completeness, format compliance, keyed-answer accuracy, output features, and response time. Two ophthalmology experts rated six domains using 5-point Likert scales, and Spearman analyses examined associations among text features, response time, and expert ratings. Results: All models achieved 100.0% initial structural acceptability, structural completeness, and format compliance. Keyed-answer accuracy was highest for ChatGPT-5.4 and Gemini 3 (both 100.0%), followed by DeepSeek (98.3%) and Kimi K2.5 and Doubao (both 95.0%). Significant between-model differences were observed across all output features and response time (all p < 0.001). ChatGPT-5.4 generated the shortest stems, Gemini 3 the shortest explanations and fastest responses, and Kimi K2.5 and Doubao the longest explanations and total outputs. Inter-rater agreement was good (ICC range, 0.835–0.885). Significant differences were found in clarity, distractor quality, and mean subjective score (all p < 0.001), but not in content rigor, educational usefulness, cognitive-level alignment, or overall usability. DeepSeek achieved the highest median mean score, while direct usability was highest for ChatGPT-5.4 (91.7%) and Gemini 3 (90.0%). Content rigor was strongly associated with overall usability ( ρ = 0.85, p < 0.05), whereas distractor quality was negatively associated with explanation length ( ρ = −0.43, p < 0.05) and total output length ( ρ = −0.37, p < 0.05). Conclusion: LLMs can reliably generate structurally valid HM-related MCQs under standardized Chinese prompting conditions. Their value may lie in supporting digital ophthalmic education and public health training, although expert oversight remains necessary because meaningful differences persist in factual accuracy, distractor quality, and direct usability.