A universal artificial intelligence platform for collaborative management of cataracts. Issue 1 (October 2019)

• Record Type: Journal Article
• Authors: Wu, Xiaohang; Huang, Yelin; Liu, Zhenzhen; Lai, Weiyi; Long, Erping; Zhang, Kai; Jiang, Jiewei; Lin, Duoru; Chen, Kexin; Yu, Tongyong; Wu, Dongxuan; Li, Cong; Chen, Yanyi; Zou, Minjie; Chen, Chuan; Zhu, Yi; Guo, Chong; Zhang, Xiayin; Wang, Ruixin; Yang, Yahan; Xiang, Yifan; Chen, Lijian; Liu, Congxin; Xiong, Jianhao; Ge, Zongyuan; Wang, Dingding; Xu, Guihua; Du, Shaolin; Xiao, Chi; Wu, Jianghao; Zhu, Ke; Nie, Danyao; Xu, Fan; Lv, Jian; Chen, Weirong; Liu, Yizhi; Lin, Haotian;
• Is Part Of: Lancet. Volume 394(2019)Special Issue 1
• Page Start: S22
• Publication Date: 2019-10
• Subjects: Medicine -- Periodicals; Medicine -- Periodicals; Medicine; Medicine; Electronic journals; Periodicals; 610.5
• DOI: 10.1016/S0140-6736(19)32358-X
• Languages: English
• ISSNs: 0140-6736

Abstract

Abstract: Background: Common diseases are not satisfactorily managed under the current health-care system because of inadequate medical resources and limited accessibility. We aimed to establish and validate a universal artificial intelligence (AI) platform for collaborative management of cataracts involving multilevel clinical scenarios, and explored an AI-based medical referral pattern to improve collaborative efficiency and resource coverage. Methods: The training and validation datasets were derived from the Chinese Medical Alliance for Artificial Intelligence, covering multilevel health-care facilities and capture modes. The datasets were labeled using a three-step strategy: capture mode recognition (modes: mydriatic-diffuse, mydriatic-slit lamp, non-mydriatic-diffuse, and nonmydriatic-slit lamp); cataract diagnosis as a normal lens, cataract, or a postoperative eye; and detection of referable cataracts with respect to cause and severity. Area under curve [AUC] was measured at each stage. We also integrated the above cataract AI agent with a real-world multilevel referral pattern involving self-monitoring at home, primary health care, and specialised hospital services. The diagnostic accuracy, treatment referral, and ophthalmologist-to-population service ratio were used to evaluate the performance and efficacy of the system. Findings: The universal AI platform and multilevel collaborative pattern showed robust diagnostic performance in three-step tasks: capture mode recognition (AUC 99·28–99·71% for the four different capture modes), cataract diagnosis (AUC for mydriatic-slit lamp mode 99·82% [95%CI 98·93–100] for normal lens vs 99·96% [99·90–100] for cataract vs 99·93% [99·78–100] for postoperative eye, and AUCs >99% for other capture modes), and detection of referable cataracts (AUCs >91% in all tests). In the real-world tertiary referral pattern, the agent suggested 30·3% of people be referred to treatment, substantially increasing the ophthalmologist-to-population service ratio by 10·2-times compared with the traditional pattern. Interpretation: The universal AI platform and multilevel collaborative pattern showed robust diagnostic performance and effective service for cataracts. The context of our AI-based medical referral pattern will be extended to other common disease conditions and resource-intensive situations. Funding: National Key Research and Development Program, National Natural Science Foundation of China, Science Foundation of China for Excellent Young Scientists, Guangdong Provincial Natural Science Foundation, Guangdong Province Universities and Colleges Youth Pearl River Scholar Funding Scheme, Science and Technology Planning Projects of Guangdong Province, Clinical Research and Translational Medical Center of Pediatric Cataract in Guangzhou City, Outstanding Young Teacher Cultivation Projects in Guangdong Province, Fundamental Research Funds for the Central Universities.