胃癌外科临床数据挖掘系统的构建与实现

Development and implementation of a clinical data mining system for gastric cancer surgery

目的 开发一款符合国际标准和临床应用习惯、适合回顾性及前瞻性临床研究、科学合理地进行肿瘤外科临床数据管理并能进行临床数据挖掘的软件系统,以满足胃癌外科多中心、大样本临床数据分析需要.方法 结合日本胃癌规约（JGCA）第13、14版及UICC分期系统,确定临床核心数据项目,根据数据挖掘理论和临床工作思路合理设计数据结构和控制逻辑,采用Microsoft Visual Basic、Vista DB等进行编程.结果 历时近1年的开发和完善,南方医科大学南方医院使用该软件系统回顾性录入600余例胃癌临床数据资料,建立起包含近4000项临床指标的、内置数据挖掘功能的数据库系统,各项功能指标经实践检验达到设计要求.该软件作为中国腹腔镜胃癌外科研究组的数据平台,已在全国内地和香港地区30家医院参与的第一期临床研究项目＂腹腔镜胃癌手术可行性的回顾性多中心研究＂中成功运行.数据挖掘功能达到预期目标,能可视化地进行复杂搜索,统计分析功能可对数据进行描述性分析.记录和数据的导入、导出满足了进行交流的需要,导出数据格式与所有统计软件兼容,无数据传输错误.结论 本软件系统在胃癌外科临床研究领域具有广泛的应用前景和网络化升级扩展的潜能.

Objective To develop a clinical database system of gastric cancer surgery integrated with data mining function for better management of clinical data and better performance of both retrospective and prospective studies. Methods Core fields for clinical data were determined based on the JGCA （13th and 14th edition） and UICC gastric cancer staging system. Microsoft Visual Basic and VistaDB were used for programming. The database structure wa6 designed according to data mining theory and clinical workflow. Results After one year of development and refinement, data of over 600 patients from our hospital were retrospectively entered, and function tests were satisfactory. This system was accepted as the database platform for the Chinese Laparoscopic Gastrointestinal Surgery Study Group （CLASS） and was successfully used in the first stage of the Multicenter Retrospective Study of the Feasibility of Laparoscopy for Gastric Cancer among 30 hospitals from both China's Mainland and Hong Kong. The data mining function met the requirements, which could carry out complex search with visualized presentation. Descriptive analyses could be performed with the analysis function. Efficient communication among institutions could be executed by data import and export with excellent compatibility and without errors. Conclusion The system has established a clinical database of approximately 4000 fields with data mining function. This system can be widely applied for the clinical research for gastric cancer.