通讯作者: 齐国强, qiguoqiang@zju.edu.cn

DOI：10.12201/bmr.202511.00056

Construction and Application of a Pediatric Kawasaki Disease Specialized Platform Based on Multimodal Data

• 摘要：目的/意义 为解决儿童川崎病（KD）临床管理中面临的早期诊断困难、治疗不规范、数据利用不足等挑战，本研究旨在构建一个基于多模态数据的川崎病专病平台，以整合与治理分散的医疗数据，支持临床诊疗优化与科研探索。方法/过程 以浙江大学医学院附属儿童医院2018-2024年间的川崎病患儿诊疗数据为基础，通过ETL技术汇聚多源异构数据，并利用自然语言处理（NLP）、机器视觉和数据治理技术进行清洗、标准化、结构化处理，建立以患儿主索引（EMPI）为核心的专病数据库。平台集成患儿管理、智能搜索、科研项目与随访管理等功能模块，并应用机器学习算法支持数据分析。结果/结论 成功构建了包含10,689例确诊患儿、覆盖18个模块共300个数据项的川崎病专病库。该平台实现了多模态数据的有效整合与质控，提供了高效的病例检索、队列构建及科研分析能力，显著提升了数据利用效率和科研敏捷性。基于多模态数据构建的川崎病专病平台能有效打破数据孤岛，形成高质量的科研数据资产，不仅为临床诊疗和病程管理提供有力支持，也为开展大样本临床研究、驱动科研创新奠定了基础，显著增强了医院在该领域的核心竞争力。

  关键词： 儿童川崎病; 专病平台; 多模态数据; 数据质控

   

  Abstract: Purpose/Significance:To address challenges in the clinical management of Kawasaki disease (KD) in children, such as difficulties in early diagnosis, non-standardized treatment, and underutilization of data, this study aimed to construct a specialized KD platform based on multimodal data to integrate and manage dispersed medical data, supporting optimized clinical diagnosis/treatment and scientific research.Method/Process: Based on the diagnostic and treatment data of KD children from the Childrens Hospital Zhejiang University School of Medicine (2018-2024), multi-source heterogeneous data were aggregated through ETL technology. Data were cleaned, standardized, and structured using Natural Language Processing (NLP), machine vision, and data governance techniques to establish a specialized database centered around the Enterprise Master Patient Index (EMPI). The platform integrated functional modules for patient management, intelligent search, research project, and follow-up management, and employed machine learning algorithms to support data analysis.Result/Conclusion: A specialized KD database was successfully constructed, containing 10,689 confirmed cases and covering 300 data items across 18 modules. The platform achieved effective integration and quality control of multimodal data, providing efficient case retrieval, cohort building, and research analysis capabilities, significantly improving data utilization efficiency and research agility. The multimodal data-based specialized KD platform effectively breaks down data silos and creates high-quality research-ready data assets. It not only provides strong support for clinical diagnosis, treatment, and disease course management but also lays a solid foundation for conducting large-sample clinical studies and driving scientific innovation, significantly enhancing the hospitals core competitiveness in this field.

  Key words: Pediatric Kawasaki Disease; Specialized Disease Platform; Multimodal Data; Data Quality Control

  提交时间：2025-11-20

  版权声明：作者本人独立拥有该论文的版权，预印本系统仅拥有论文的永久保存权利。任何人未经允许不得重复使用。
• html

• 图表

• 龙思哲, 张敦明, 吴震天, 黎鹏安, 王毅, 周毅, 张武军. 基于数据治理的专病数据库建设实践. 2022. doi: 10.12201/bmr.202207.00012

  连万民. 基于数据治理的脑血管专病数据库建设实践. 2022. doi: 10.12201/bmr.202210.00011

  夏刚山, 张明明, 李晓惠. 血小板-白细胞聚集体在川崎病发病机制的研究进展. 2025. doi: 10.12201/bmr.202502.00058

  毛 丹, 夏 天, 刘捷宸, 张 诚. 基于疾控综合管理平台的数据监管业务应用实践. 2025. doi: 10.12201/bmr.202503.00041

  刘迷迷, 杜国霞, 周毅, 吴斌, 吴晓瑛. 肝硬化专病数据库建设与应用研究. 2021. doi: 10.12201/bmr.202104.00014

  张亚男. 基于NLP构建病历后结构化专病数据库的探索和实践. 2024. doi: 10.12201/bmr.202406.00007

  唐蓉, 张华东, 丁贤彬, 夏雨逸, 江玲. 基于医疗大数据的区域慢性病监测管理信息平台构建与应用实践. 2025. doi: 10.12201/bmr.202506.00071

  王耀国, 唐诗诗, 刘泓泽, 安雨婷, 周毅. 基于本地大模型的骨质疏松专病数据库构建流程优化研究. 2024. doi: 10.12201/bmr.202410.00002

  张亚, .何凯. 川崎病病因与发病机制的研究进展. 2025. doi: 10.12201/bmr.202510.00016

  吕慧, 初平, 张柠, 徐新, 赵成松. 基于真实世界数据的儿科处方流转平台用户特征比较研究. 2024. doi: 10.12201/bmr.202404.00017

• 序号	提交日期	编号	操作
  1	2025-09-10	10.12201/bmr.202511.00056V1	下载

• 公开评论  匿名评论  仅发给作者 提交评论