人工智能数据采集在慢性乙型肝炎患者真实世界研究中的应用  

Application of artificial intelligence algorithms for data collection in real-world studies of chronic hepatitis B patients

作　　者：周晓梅 曾涛 廖莹颖 张一博 李青海 Jaime Smith 张麟 王超 崇雨田[2] 李新华[2] ZHOU Xiaomei;ZENG Tao;LIAO Yingying;ZHANG Yibo;LI Qinghai;SMITH Jaime;ZHANG Lin;WANG Chao;CHONG Yutian;LI Xinhua(Information Center,the Third Affiliated Hospital of Sun Yat-sen University,Guangzhou 510630,Guangdong,China;Department of Infectious Diseases,Key Laboratory of Liver Disease of Guangdong Province,the Third Affiliated Hospital of Sun Yat-Sen University,Guangzhou 510630,Guangdong,China;KingPoint Data Technology Co.,Ltd.Guangzhou 510635,Guangdong,China;PAREXEL.China Co.,Ltd.Shanghai 200120,China)

机构地区：[1]中山大学附属第三医院信息中心,广东广州510630 [2]中山大学附属第三医院感染性疾病科∥广东省肝病重点实验室,广东广州510630 [3]广东精点数据科技股份有限公司,广东广州510635 [4]精鼎医药研究开发(上海)有限公司,上海200120

出　　处：《暨南大学学报（自然科学与医学版）》2024年第1期77-83,共7页Journal of Jinan University(Natural Science & Medicine Edition)

基　　金：广东省重点领域研发计划项目(2019B020228001)。

摘　　要：目的:开发一套慢性乙型肝炎(乙肝)的人工智能(AI)数据采集工具,以解决传统的多中心数据采集效率低下的问题。方法:基于国际通用的数据标准,将AI的文字识别和自然语言处理应用于慢性乙肝真实世界队列研究数据采集,实现多种格式数据(包括图片格式的原始数据)的电子化采集、结构化处理,随后自动将数据填入研究电子数据采集(REDCap)系统中的电子病历报告表(eCRF)。结果:AI工具辅助数据采集与纯人工数据采集具有相同的平均准确率,均达到98.66%(P=0.23),但前者所需时间减少75.49%(P<0.05)。结论:本研究开发的AI数据采集工具可显著提高研究数据采集效率,为真实世界研究数据采集提供了新的模式。Objective:To develop an artificial intelligence(AI)algorithms tool for data collection in chronic hepatitis B,addressing the issue of low efficiency in traditional multicenter data collection.Methods:Based on internationally recognized data standards,this study applies mature AI techniques in the field of computer science,such as optical character recognition and natural language processing,to the data collection of real-world cohorts in chronic hepatitis B research.This tool enables the electronic collection of various data formats,including digitization of raw data in image format,structured processing,and automatic population of data into electronic case report forms(eCRFs)designed in the Research Electronic Data Capture(REDCap)system.Results:Experimental results demonstrate that the AI algorithmstool for data collection achieves the same average accuracy as manual data collection(P=0.23),with an accuracy rate of 98.66%,while reducing the time required by 75.49%(P<0.05)compared to manual collection.Conclusion:The AI algorithms tool for data collection developed in this study significantly improves the efficiency of research data collection and brings new developments to the real-world research data collection paradigm.

关 键 词：数据采集 慢性乙型肝炎 人工智能(AI) 自然语言处理 文字识别 

分 类 号：R512.6[医药卫生—内科学]