Aslakson Eric, Szekely Smadar, Vernon Suzanne D, Bateman Lucinda, Baumbach Jan, Setty Yaki
Poiema, LLC, 375 Chelsea Cir NE, Atlanta, GA 30307, USA.
Theor Biol Med Model. 2012 Jun 15;9:22. doi: 10.1186/1742-4682-9-22.
Medical records accumulate data concerning patient health and the natural history of disease progression. However, methods to mine information systematically in a form other than an electronic health record are not yet available. The purpose of this study was to develop an object modeling technique as a first step towards a formal database of medical records.
Live Sequence Charts (LSC) were used to formalize the narrative text obtained during a patient interview. LSCs utilize a visual scenario-based programming language to build object models. LSC extends the classical language of UML message sequence charts (MSC), predominantly through addition of modalities and providing executable semantics. Inter-object scenarios were defined to specify natural history event interactions and different scenarios in the narrative text.
A simulated medical record was specified into LSC formalism by translating the text into an object model that comprised a set of entities and events. The entities described the participating components (i.e., doctor, patient and record) and the events described the interactions between elements. A conceptual model is presented to illustrate the approach. An object model was generated from data extracted from an actual new patient interview, where the individual was eventually diagnosed as suffering from Chronic Fatigue Syndrome (CFS). This yielded a preliminary formal designated vocabulary for CFS development that provided a basis for future formalism of these records.
Translation of medical records into object models created the basis for a formal database of the patient narrative that temporally depicts the events preceding disease, the diagnosis and treatment approach. The LSCs object model of the medical narrative provided an intuitive, visual representation of the natural history of the patient's disease.
医疗记录积累了有关患者健康和疾病进展自然史的数据。然而，尚未有以电子健康记录以外的形式系统挖掘信息的方法。本研究的目的是开发一种对象建模技术，作为迈向正式医疗记录数据库的第一步。
使用活序列图（LSC）将患者访谈期间获得的叙述文本形式化。LSC利用基于可视化场景的编程语言来构建对象模型。LSC扩展了统一建模语言消息序列图（MSC）的经典语言，主要是通过添加模态并提供可执行语义。定义了对象间场景，以指定叙述文本中的自然史事件交互和不同场景。
通过将文本转换为包含一组实体和事件的对象模型，将模拟医疗记录指定为LSC形式。实体描述了参与组件（即医生、患者和记录），事件描述了元素之间的交互。提出了一个概念模型来说明该方法。从实际新患者访谈中提取的数据生成了一个对象模型，该患者最终被诊断为患有慢性疲劳综合征（CFS）。这产生了一个用于CFS发展的初步正式指定词汇表，为这些记录的未来形式化提供了基础。
将医疗记录转换为对象模型为患者叙述的正式数据库奠定了基础，该数据库在时间上描绘了疾病之前的事件、诊断和治疗方法。医疗叙述的LSCs对象模型提供了患者疾病自然史的直观视觉表示。
