%0 Journal Article
%T 人工智能辅助基层糖尿病视网膜病变筛防实践<br>Practice of Artificial Intelligence-Assisted Screening and Prevention of Diabetic Retinopathy at the Grass-Roots Level
%A 王少鹏
%A 刘延晶
%A 刘娜
%A 刘红霞
%A 高丽芬
%A 翟改霞
%A 路晖
%J Advances in Clinical Medicine
%P 2380-2386
%@ 2161-8720
%D 2020
%I Hans Publishing
%R 10.12677/ACM.2020.1010359
%X <div style=\"text-align:justify;\">
	背景：人工智能辅助眼底筛查作为一种新兴的眼科检查模式，可以弥补专科医师不足和患者可及性差等诸多问题。目的：本研究以地区中心医院医联体为载体，探索基层糖尿病视网膜病变(糖网)筛防新模式。方法：通过医联体系统从基层医疗机构纳入2300名患者，完成眼科检查并经AI初判、人工复核形成眼底诊断结论。筛查以老龄人口为主的社区糖网患病情况。结果：社区需转诊糖尿病视网膜病变患者AI检出的灵敏度84.67%、特异度93.87%。全部受检人群糖网患病率11.43%；糖尿病患者糖网患病率24.70%，需转诊糖网患病率19.65%；无糖尿病病史人群糖网患病率7.01%，需转诊糖网患病率5.10%。结论：人工智能辅助基层老年人群糖网普查效率高、诊断精度好，有助于糖尿病患者及时发现威胁视力的视网膜并发症，也有助于糖尿病的筛查和及时管理。<br />
Background: Artificial intelligence-assisted fundus screening as a novel ophthalmological examination can solve many problems such as the primary shortage of ophthalmologists and patient's accessibility for health services. Objective: The urban central hospital built a medical union with a number of town and township central hospitals to execute a novel screening method for diabetic retinopathy at primary medical care. Methods: A total of 2300 elderly patients in communities were enrolled from primary medical hospital through the medical union system, whose retinagraphy was taken and the fundus diagnosis conclusion was formed by AI and reviewed by center hospital ophthalmologists. Results: The sensitivity and specificity of AI diagnosis of referral diabetic retinopathy in communities were 84.67% and 93.87%, respectively. The prevalence rate of diabetic retinopathy was 11.43% in all population. The prevalence rate of diabetic retinopathy in patients with diabetes was 24.70% and the prevalence rate of diabetic retinopathy which needs referral in these people is 19.65%. The prevalence rate of diabetic retinopathy in people with no history of diabetes is 7.01%, and the prevalence rate of diabetic retinopathy which needs referral in these people is 5.10%. Conclusion: Artificial intelligence-assisted diabetic retinopathy screening of the elderly in communities has high efficiency and good diagnostic accuracy, which is helpful for diabetic patients to detect retinal complications that threaten vision in time, as well as for diabetes timely management.
</div>
%K 人工智能，糖尿病视网膜病，基层<br>Artificial Intelligence
%K Diabetic Retinopathy
%K Grass-Roots Level
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=38270