Scholarlink

<p>Diabetic retinopathy (DR) and diabetic macular edema (DME) are leading causes of vision impairment worldwide, necessitating continuous monitoring and timely interventions due to their chronic, progressive nature. This systematic review evaluates the effectiveness, limitations, and economic benefits of adaptive AI models in managing DR/DME (2020–2025 (February)), focusing on their potential to transform static diagnostics into dynamic, personalized care. We analyzed 119 studies following PRISMA guidelines, prioritizing peer-reviewed research on AI models for dynamic monitoring, personalized risk stratification, and teleophthalmology. Key databases included PubMed, IEEE Xplore, and Scopus. Adaptive AI models demonstrated significant clinical and economic impact. Hybrid models integrating electronic health records (EHRs) and genomic data improved DR progression prediction (AUC: 0.93), while LSTM networks reduced treatment costs by 20% through optimized anti-VEGF therapy. Vision Transformers achieved superior accuracy in real-time monitoring (AUC: 0.95), and teleophthalmology systems like EyePACS reduced unnecessary referrals by 30% in underserved regions. Key limitations include inconsistent real-world validation, regulatory delays (e.g., &lt;5% of models are FDA-approved), and biases in generalization across diverse populations. Adaptive AI models show transformative potential for DR and DME management but require robust validation, ethical frameworks, and interoperability improvements. Therefore, to fully realize the transformative potential of adaptive AI models for DR and DME management, future research must prioritize robust validation, ethical frameworks, and interoperability improvements, focusing on federated learning for data privacy, edge computing for real-time analysis, and integration of socioeconomic factors to enhance equity.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Revolutionizing diabetic retinopathy and macular edema management: a systematic review on the transformative potential of artificial intelligence

  • 期刊论文

  • 发表日期：

    2025年6月10日

查看全文
  • Saba Ghazanfar Ali,
  • Xiaolong Yang,
  • Saleha Masood,
  • Zainab Ghazanfar,
  • Younhyun Jung,
  • Tingli Chen,
  • Xiangning Wang

摘要

Diabetic retinopathy (DR) and diabetic macular edema (DME) are leading causes of vision impairment worldwide, necessitating continuous monitoring and timely interventions due to their chronic, progressive nature. This systematic review evaluates the effectiveness, limitations, and economic benefits of adaptive AI models in managing DR/DME (2020–2025 (February)), focusing on their potential to transform static diagnostics into dynamic, personalized care. We analyzed 119 studies following PRISMA guidelines, prioritizing peer-reviewed research on AI models for dynamic monitoring, personalized risk stratification, and teleophthalmology. Key databases included PubMed, IEEE Xplore, and Scopus. Adaptive AI models demonstrated significant clinical and economic impact. Hybrid models integrating electronic health records (EHRs) and genomic data improved DR progression prediction (AUC: 0.93), while LSTM networks reduced treatment costs by 20% through optimized anti-VEGF therapy. Vision Transformers achieved superior accuracy in real-time monitoring (AUC: 0.95), and teleophthalmology systems like EyePACS reduced unnecessary referrals by 30% in underserved regions. Key limitations include inconsistent real-world validation, regulatory delays (e.g., <5% of models are FDA-approved), and biases in generalization across diverse populations. Adaptive AI models show transformative potential for DR and DME management but require robust validation, ethical frameworks, and interoperability improvements. Therefore, to fully realize the transformative potential of adaptive AI models for DR and DME management, future research must prioritize robust validation, ethical frameworks, and interoperability improvements, focusing on federated learning for data privacy, edge computing for real-time analysis, and integration of socioeconomic factors to enhance equity.