Ehrler Frédéric, Ruch Patrick, Geissbuhler Antoine, Lovis Christian
Artificial Intelligence laboratory, University of Geneva, Geneva, Switzerland.
Stud Health Technol Inform. 2007;129(Pt 1):417-21.
Patient records contain most crucial documents for managing the treatments and healthcare of patients in the hospital. Retrieving information from these records in an easy, quick and safe way helps care providers to save time and find important facts about their patient's health. This paper presents the scalability issues induced by the indexing and the retrieval of the information contained in the patient records. For this study, EasyIR, an information retrieval tool performing full text queries and retrieving the related documents has been used. An evaluation of the performance reveals that the indexing process suffers from overhead consequence of the particular structure of the patient records. Most IR tools are designed to manage very large numbers of documents in a single index whereas in our hypothesis, one index per record, which usually implies few documents, has been imposed. As the number of modifications and creations of patient records are significant in a day, using a specialized and efficient indexation tool is required.
患者记录包含医院中管理患者治疗和医疗保健的最关键文件。以简单、快速和安全的方式从这些记录中检索信息有助于护理人员节省时间并找到有关患者健康的重要事实。本文介绍了患者记录中信息索引和检索所引发的可扩展性问题。对于本研究，使用了EasyIR，这是一种执行全文查询并检索相关文档的信息检索工具。性能评估表明，索引过程受到患者记录特定结构带来的开销影响。大多数信息检索工具旨在在单个索引中管理大量文档，而在我们的假设中，每条记录一个索引，这通常意味着文档数量较少。由于患者记录的修改和创建数量在一天中很可观，因此需要使用专门且高效的索引工具。
