Semantic Web-Based Medical Documentation Processing for Healthcare Collaboration

目前的医疗保健是一个不断上升的成本，加上人口老龄化和日益复杂的并发慢性医疗疾病。为这些患者提供的护理是由大型跨学科团队进行的，他们面临的挑战是同步了解每个成员的观点，并制定连贯和安全的治疗计划。与此环境相关的挑战部分存在于如何编写文档。绝大多数文档都是由竖井中的非结构化/半结构化文本组成的。为了充分了解团队的观点，团队成员必须在能够为治疗计划提供综合贡献之前检查每个单独的条目。因此，如果有人同意医学文档是临床印象和计划的核心，如果计算机可以(i)正确解析和注释医学文档；（ii）从各种电子来源的非结构化文档中构建高语义结构，以及（iii）预测临床环境。然后，计算机可以采取哪些行动，使患者、医疗团队和各级决策者受益？本研究计划通过研究医疗文档处理和分析的新方法来解决上述挑战，从而使文档不再对信息技术惰性，而是知识获取的丰富来源和驱动广泛的软件流程的路线图，范围从协作治疗计划到及时决策支持。为了实现这个级别的智能，我们采用了一种新的方法来结合模型驱动软件开发（MDSD）、本体工程和上下文处理的优点来解决(i)、（ii）和（iii）。

The present landscape of healthcare is one of escalating costs coupled with an aging population with increasing complexity in concurrent chronic medical illnesses. Care provisioned to these patients are carried out by large interdisciplinary teams that are challenged with being synchronously aware of each member's view points and generating coherent and safe treatment plans. Part of challenges relating to this environment exists in how documentation occurs. The vast majority of documentation consists of unstructured/semi-structured text that occurs in silos. In order to fully understand the team's point of view, a team member would have to inspect every individual entry before being able to provide an integrated contribution to the treatment plan. Therefore, if one agrees that within medical documentation lies the very heart of clinical impressions and planning, what if a computer can (i) correctly parse and annotate medical documentation; (ii) build high semantic structures out of unstructured documents from various electronic sources, and (iii) predict clinical contexts. What actions could then a computer take that benefit the patient, the healthcare team, and all levels of decision makers? This research program addresses the above challenges by investigating a novel approach to medical documentation processing and analysis, whereby the document is no longer inert to information technology, but a rich source for knowledge acquisition and a road map to drive a wide range of software processes which range from collaborative treatment planning to just-in-time decision support. To accomplish this level of intelligence, we adopt a new approach to combining benefits of Model Driven Software Development (MDSD), ontology engineering, and context processing to solve (i), (ii) and (iii).