Deshpande Aniruddha M, Brandt Cynthia, Nadkarni Prakash M
Center for Medical Informatics, Yale University School of Medicine, PO Box 208009, New Haven, CT 06520-8009, USA.
Int J Med Inform. 2003 Apr;70(1):59-77. doi: 10.1016/s1386-5056(02)00183-1.
We describe an interface and architecture for ad hoc temporal query of TrialDB, a clinical study data management system (CSDMS). A clinical study focuses primarily on the effect of therapy on a group of patients, who have individually enrolled in a study at different times. Relative times (chronological offsets from the time of enrollment) are therefore more useful than absolute times when collectively describing therapeutic or adverse events. For logistic reasons, study parameter values are typically recorded at fixed relative times ('study events'), which serve as time-stamps and can be used by CSDMS temporal query algorithms to simplify temporal computations. The entity-attribute-value model of clinical data storage, used by both CSDMSs and clinical patient record systems, complicates temporal query. To apply temporal operators, data for parameters of interest must first be transiently converted into conventional relational form, with one column per parameter.
我们描述了一种用于临床研究数据管理系统（CSDMS）TrialDB的临时时间查询的接口和架构。临床研究主要关注治疗对一组患者的效果，这些患者在不同时间分别参加了一项研究。因此，在集体描述治疗或不良事件时，相对时间（从入组时间起的时间偏移）比绝对时间更有用。出于逻辑原因，研究参数值通常在固定的相对时间（“研究事件”）记录，这些时间用作时间戳，并且可被CSDMS时间查询算法用于简化时间计算。临床数据存储的实体-属性-值模型，被CSDMS和临床患者记录系统所使用，使时间查询变得复杂。为了应用时间运算符，感兴趣参数的数据必须首先临时转换为传统的关系形式，每个参数一列。
