Bache Richard, Daniel Christel, James Julie, Hussain Sajjad, McGilchrist Mark, Delaney Brendan, Taweel Adel
Dept. of Informatics, School of Natural and Mathematical Sciences, King's College London, UK.
Assistance Publique - Hôpitaux de Paris & INSERM, U1142, Paris, France.
Stud Health Technol Inform. 2014;205:273-7.
The HL7 RIM (Reference Information Model) is a commonly used standard for the exchange of clinical data and can be employed for integrating the patient care and clinical research domains. Yet it is not sufficiently well specified to ensure a canonical representation of structured clinical data when used for the automated evaluation of eligibility criteria from a clinical trial protocol. We present an approach to further constrain the RIM to create a common information model to hold clinical data. In order to demonstrate our approach, we identified 132 distinct data elements from 10 rich clinical trails. We then defined a taxonomy to (i) identify the types of data elements that would need to be stored and (ii) define the types of predicate that would be used to evaluate them. This informed the definition of a pattern used to represent the data, which was shown to be sufficient for storing and evaluating the clinical statements required by the trials.
HL7参考信息模型（RIM）是临床数据交换中常用的标准，可用于整合患者护理和临床研究领域。然而，当用于从临床试验方案自动评估入选标准时，它的规范还不够完善，无法确保结构化临床数据的规范表示。我们提出了一种进一步约束RIM的方法，以创建一个用于保存临床数据的通用信息模型。为了演示我们的方法，我们从10个丰富的临床试验中识别出132个不同的数据元素。然后，我们定义了一种分类法，用于（i）识别需要存储的数据元素类型，以及（ii）定义用于评估这些数据元素的谓词类型。这为表示数据的模式定义提供了依据，该模式被证明足以存储和评估试验所需的临床声明。
