Aim To develop a deep learning (DL) model that predicts age from fundus images (retinal age) and to investigate the association between retinal age gap (retinal age predicted by DL model minus chronological age) and mortality risk. Methods A total of 80 169 fundus images taken from 46 969 participants in the UK Biobank with reasonable quality were included in this study. Of these, 19 200 fundus images from 11 052 participants without prior medical history at the baseline examination were used to train and validate the DL model for age prediction using fivefold cross-validation. A total of 35 913 of the remaining 35 917 participants had available mortality data and were used to investigate the association between retinal age gap and mortality. Results The DL model achieved a strong correlation of 0.81 (p<0·001) between retinal age and chronological age, and an overall mean absolute error of 3.55 years. Cox regression models showed that each 1 year increase in the retinal age gap was associated with a 2% increase in risk of all-cause mortality (hazard ratio (HR)=1.02, 95% CI 1.00 to 1.03, p=0.020) and a 3% increase in risk of cause-specific mortality attributable to non-cardiovascular and non-cancer disease (HR=1.03, 95% CI 1.00 to 1.05, p=0.041) after multivariable adjustments. No significant association was identified between retinal age gap and cardiovascular- or cancer-related mortality. Conclusions Our findings indicate that retinal age gap might be a potential biomarker of ageing that is closely related to risk of mortality, implying the potential of retinal image as a screening tool for risk stratification and delivery of tailored interventions.

Background: Large language models (LLMs) demonstrated advanced performance in processing clinical information. However, commercially available LLMs lack specialized medical knowledge and remain susceptible to generating inaccurate information. Given the need for self-management in diabetes, patients commonly seek information online. We introduce the Retrieval-augmented Information System for Enhancement (RISE) framework and evaluate its performance in enhancing LLMs to provide accurate responses to diabetes-related inquiries. Objective: This study aimed to evaluate the potential of the RISE framework, an information retrieval and augmentation tool, to improve the LLM’s performance to accurately and safely respond to diabetes-related inquiries. Methods: The RISE, an innovative retrieval augmentation framework, comprises 4 steps: rewriting query, information retrieval, summarization, and execution. Using a set of 43 common diabetes-related questions, we evaluated 3 base LLMs (GPT-4, Anthropic Claude 2, Google Bard) and their RISE-enhanced versions respectively. Assessments were conducted by clinicians for accuracy and comprehensiveness and by patients for understandability. Results: The integration of RISE significantly improved the accuracy and comprehensiveness of responses from all 3 base LLMs. On average, the percentage of accurate responses increased by 12% (15/129) with RISE. Specifically, the rates of accurate responses increased by 7% (3/43) for GPT-4, 19% (8/43) for Claude 2, and 9% (4/43) for Google Bard. The framework also enhanced response comprehensiveness, with mean scores improving by 0.44 (SD 0.10). Understandability was also enhanced by 0.19 (SD 0.13) on average. Data collection was conducted from September 30, 2023 to February 5, 2024. Conclusions: The RISE significantly improves LLMs’ performance in responding to diabetes-related inquiries, enhancing accuracy, comprehensiveness, and understandability. These improvements have crucial implications for RISE’s future role in patient education and chronic illness self-management, which contributes to relieving medical resource pressures and raising public awareness of medical knowledge.

Fundus fluorescein angiography (FFA) is a crucial diagnostic tool for chorioretinal diseases, but its interpretation requires significant expertise and time. Prior studies have used Artificial Intelligence (AI)-based systems to assist FFA interpretation, but these systems lack user interaction and comprehensive evaluation by ophthalmologists. Here, we used large language models (LLMs) to develop an automated interpretation pipeline for both report generation and medical question-answering (QA) for FFA images. The pipeline comprises two parts: an image-text alignment module (Bootstrapping Language-Image Pre-training) for report generation and an LLM (Llama 2) for interactive QA. The model was developed using 654,343 FFA images with 9392 reports. It was evaluated both automatically, using language-based and classification-based metrics, and manually by three experienced ophthalmologists. The automatic evaluation of the generated reports demonstrated that the system can generate coherent and comprehensible free-text reports, achieving a BERTScore of 0.70 and F1 scores ranging from 0.64 to 0.82 for detecting top-5 retinal conditions. The manual evaluation revealed acceptable accuracy (68.3%, Kappa 0.746) and completeness (62.3%, Kappa 0.739) of the generated reports. The generated free-form answers were evaluated manually, with the majority meeting the ophthalmologists’ criteria (error-free: 70.7%, complete: 84.0%, harmless: 93.7%, satisfied: 65.3%, Kappa: 0.762–0.834). This study introduces an innovative framework that combines multi-modal transformers and LLMs, enhancing ophthalmic image interpretation, and facilitating interactive communications during medical consultation.

Early detection of eye diseases is vital for preventing vision loss. Existing ophthalmic artificial intelligence models focus on single modalities, overlooking multi-view information and struggling with rare diseases due to long-tail distributions. We propose EyeCLIP, a multimodal visual-language foundation model trained on 2.77 million ophthalmology images from 11 modalities with partial clinical text. Our novel pretraining strategy combines self-supervised reconstruction, multimodal image contrastive learning, and image-text contrastive learning to capture shared representations across modalities. EyeCLIP demonstrates robust performance across 14 benchmark datasets, excelling in disease classification, visual question answering, and cross-modal retrieval. It also exhibits strong few-shot and zero-shot capabilities, enabling accurate predictions in real-world, long-tail scenarios. EyeCLIP offers significant potential for detecting both ocular and systemic diseases, and bridging gaps in real-world clinical applications.

Background: Retinal parameters could reflect systemic vascular changes. With the advances of deep learning technology, we have recently developed an algorithm to predict retinal age based on fundus images, which could be a novel biomarker for aging and mortality. Therefore, we aim to investigate associations of retinal age gap with arterial stiffness index and incident cardiovascular disease (CVD). Methods: A deep learning model was trained based on 19 200 fundus images of 11 052 participants without any medical history at baseline to predict the retinal age. Retinal age gap (retinal age predicted minus chronological age) was generated for the remaining 35 917 participants. Regression models were used to assess the association between retinal age gap and arterial stiffness index. Cox proportional hazards regression models and restricted cubic splines were used to explore the association between retinal age gap and incident CVD. Results: We found each 1-year increase in retinal age gap was associated with increased arterial stiffness index (β=0.002 [95% CI, 0.001–0.003]; P <0.001). After a median follow-up of 5.83 years (interquartile range: 5.73–5.97), 675 (2.00%) developed CVD. In the fully adjusted model, each 1-year increase in retinal age gap was associated with a 3% increase in the risk of incident CVD (hazard ratio=1.03 [95% CI, 1.01–1.06]; P =0.014). In the restricted cubic splines analysis, the risk of incident CVD increased significantly when retinal age gap reached 1.21 (hazard ratio=1.05 [95% CI, 1.00–1.10]; P -overall <0.0001; P -nonlinear=0.0681). Conclusions: We found that retinal age gap was significantly associated with arterial stiffness index and incident CVD events, supporting the potential of this novel biomarker in identifying individuals at high risk of future CVD events.

To evaluate the accuracy and reasoning ability of DeepSeek-R1 and three recently released large language models (LLMs) in bilingual complex ophthalmology cases. A total of 130 multiple-choice questions (MCQs) related to diagnosis (n ​= ​39) and management (n ​= ​91) were collected from the Chinese ophthalmology senior professional title examination and categorized into six topics. These MCQs were translated into English. Responses from DeepSeek-R1, Gemini 2.0 Pro, OpenAI o1 and o3-mini were generated under default configurations between February 15 and February 20, 2025. Accuracy was calculated as the proportion of correctly answered questions, with omissions and extra answers considered incorrect. Reasoning ability was evaluated through analyzing reasoning logic and the causes of reasoning errors. DeepSeek-R1 demonstrated the highest overall accuracy, achieving 0.862 in Chinese MCQs and 0.808 in English MCQs. Gemini 2.0 Pro, OpenAI o1, and OpenAI o3-mini attained accuracies of 0.715, 0.685, and 0.692 in Chinese MCQs (all P ​<0.001 compared with DeepSeek-R1), and 0.746 (P ​= ​0.115), 0.723 (P ​= ​0.027), and 0.577 (P ​<0.001) in English MCQs, respectively. DeepSeek-R1 achieved the highest accuracy across five topics in both Chinese and English MCQs. It also excelled in management questions conducted in Chinese (all P ​<0.05). Reasoning ability analysis showed that the four LLMs shared similar reasoning logic. Ignoring key positive history, ignoring key positive signs, misinterpretation of medical data, and overuse of non–first-line interventions were the most common causes of reasoning errors. DeepSeek-R1 demonstrated superior performance in bilingual complex ophthalmology reasoning tasks than three state-of-the-art LLMs. These findings highlight the potential of advanced LLMs to assist in clinical decision-making and suggest a framework for evaluating reasoning capabilities.