查詢結果分析
來源資料
相關文獻
- 以人工智慧篩檢糖尿病視網膜病變之檢測正確性--系統性文獻回顧
- 人工智慧於視網膜疾病之應用
- 以人工智慧決策分析糖尿病視網膜病變影像
- 人工智慧輔助診斷軟體與糖尿病視網膜病變篩檢
- 人工智慧輔助診斷糖尿病視網膜病變之成本效用分析
- Evaluation of the Pain-Relieving Effect of Carbamazepine (Tegretol[feb9]) during Panretinal Photocoagulation
- 人工智慧在軍事上之應用
- 糖尿病視網膜病變
- Detection of Diabetic Retinopathy Using Non-Mydriatic Fundus Photography at Chang Gung Memorial Hospital
- 電腦圍棋形勢判斷系統之研製
頁籤選單縮合
| 題 名 | 以人工智慧篩檢糖尿病視網膜病變之檢測正確性--系統性文獻回顧=Test Accuracy of Artificial Intelligence in Diabetic Retinopathy Screening: A Systematic Review |
|---|---|
| 作 者 | 陳圓媛; 鄭昀瑄; 連加恩; 滕欣; 蕭鈺靜; 謝佩芬; 楊惇筑; 郭玟妤; 董鈺琪; | 書刊名 | 醫務管理期刊 |
| 卷 期 | 26:2 2025.06[民114.06] |
| 頁 次 | 頁138-175 |
| 分類號 | 415.6686 |
| 關鍵詞 | 人工智慧; 糖尿病視網膜病變; 檢測正確性; Artificial intelligence; Diabetic retinopathy; Test accuracy; |
| 語 文 | 中文(Chinese) |
| DOI | 10.6174/JHM.202506_26(2).138 |
| 中文摘要 | 目的:本文欲評估人工智慧輔助診斷糖尿病視網膜病變技術的準確性。方法:利用PubMed、Web of Science、Cochrane Library與Scopus資料庫對2024年1月18日之前的相關文獻進行了系統性回顧,篩選納入54篇研究。結果:人工智慧篩檢任何類型糖尿病視網膜病變的敏感度介於69.00%至98.80%間,特異度介於74.30%至98.50%間。在篩檢可轉診糖尿病視網膜病變方面,人工智慧的敏感度介於72.08%至100.00%間,有9項研究達到100.00%;特異度介於54.00%至100.00%間。對於美國食品藥物管理局設置的優越性終點,8成多的研究人工智慧篩檢可轉診糖尿病視網膜病變敏感度大於85.00%,7成多的研究特異度大於82.50%。結論:由於人工智慧相較於人工判讀在診斷糖尿病視網膜病變方面具有較好的敏感度和特異度,醫療服務提供者可使用人工智慧篩檢糖尿病視網膜病變,以提高檢測正確性。 |
| 英文摘要 | Objectives: This review systematically evaluated the accuracy of artificial intelligence-based diagnostic techniques for Diabetic Retinopathy (DR). Methods: We collected relevant literature up to January 18, 2024, using the PubMed, Web of Science, Cochrane Library, and Scopus databases, ultimately including 54 studies. Results: The sensitivity of artificial intelligence in screening for any type of DR ranged from 69.00% to 98.80%, and the specificity ranged from 74.30% to 98.50%. For Referable Diabetic Retinopathy (RDR) screening, the sensitivity of artificial intelligence screening ranged from 72.08% to 100.00%, with 9 studies achieving 100.00%, and the specificity ranged from 54.00% to 100.00%. Regarding the superiority endpoints set by the U.S. Food and Drug Administration, over 80% of studies on AI-based screening for RDR demonstrated a sensitivity greater than 85.00%, while over 70% of studies showed a specificity greater than 82.50%. Conclusions: Due to the better sensitivity and specificity of artificial intelligence compared to humans in diagnosing diabetic retinopathy, healthcare providers can use artificial intelligence in diabetic retinopathy screening, thereby improving accuracy. |
本系統中英文摘要資訊取自各篇刊載內容。