基于Web爬虫技术的电子病历信息聚合工具的开发及验证

Development and validation of a Web-crawler-based medical records information aggregation tool

目的:以放疗科的临床和科研工作需求为导向,开发和验证一种基于Web爬虫技术的电子病历信息聚合工具。方法:基于Selenium框架和Python编程语言,设计一种基于Web爬虫的病历信息聚合工具,并列举两个实际应用场景:回顾性研究中的数据准备工作以及聚合报告新入院患者的常规临床检查结果作为例子进行说明。测试该工具对临床工作流程的益处,比较自动化方法和手动方法的效率和准确性。结果:与人工方法相比,自动信息聚合工具表现出优秀的效率和准确性。对于第一个场景,自动化工具从3541例患者中提取出110例放射性肺炎的病例,平均每例患者耗时54 s;而人工方法提取出相同数量的病例,平均每例患者耗时90 s。对于另一个例子,自动化方法平均每例患者耗时10 s,而人工方法平均每例患者耗时75 s。结论:本工作开发的工具可以在较低访问权限下实现临床和科研工作所需的数据检索、分类汇总等特殊功能,具有安全、高效、准确、跨平台、易拓展等优势。

Objective To develop and validate a Web-crawler-based medical records information aggregation tool for the clinical and research practices in Department of Radiation Oncology.Methods Based on Selenium framework and Python programming language,a Web-crawler-based medical records information aggregation tool was designed.The benefits of the proposed tool for clinical workflow were analyzed under two scenarios,namely the data preparation for retrospective study and the aggregate report of routine clinical examination results of newly-admitted patients.Moreover,the efficiency and accuracy between automatic and manual methods were also compared.Results Compared with manual method,the automatic information aggregation tool had superior efficiency and accuracy.For the first scenario,automatic method identified 110 radiation pneumonia cases out of 3541 cases at the time cost of about 54 s per case,while manual methods also identified the same number of radiation pneumonia cases but cost about 90 s per case.For the other scenario,automatic method cost about 10 s per case and manual method cost 75 s per case.Conclusion The Web-crawler-based medical records information aggregation tool can implement special functions such as data retrieval and subtotal for clinical and scientific researches under low access rights,with the advantages of security,efficiency,accuracy,cross-platform and easy-to-extend application.