INTELLIGENT CRITIQUING OF CLINICAL-GUIDELINE APPLICATION

临床指南是规范和统一提高医疗质量的有力手段。在拟议的研究项目中，我们称之为Asgaard/DeGel，我们将设计和实施软件解决方案，以支持护理提供者对临床指南应用的设计、应用，特别是质量评估。我们的长期目标包括为规范、应用和运行提供标准化的计算框架，以及临床指南的回顾性质量评估。

Clinical guidelines are a powerful method for standardization and uniform improvement of the quality of medical care. In the proposed research project, which we call Asgaard/DeGel, we will design and implement software solutions that support the design, application, and, in particular, quality assessment of the application of clinical guidelines by care providers. Our long-term objectives include provision of a standardized computational framework for specification, application, and runtime as well as retrospective quality assessment of clinical-guidelines. We will provide these solutions by using a novel framework for specification and interpretation of clinical guidelines. The new framework encodes explicitly both the therapeutic actions indicated by the guideline and the intermediate clinical process and outcome goals of the guideline designers, which we refer to as intentions. Intentions are represented as temporal patterns of care-provider actions or patient states to be maintained, achieved or avoided, using results from our previous work on interpretation and summarization of time-oriented clinical data. Intentions include both the intermediate and overall process pattern (care provider actions) to follow and the expected intermediate and final outcome pattern (patient states). The tasks that we are focusing on include (1) design-time tasks, such as specification, validation, and verification of the guideline, and (2) runtime tasks such as automated support to application of the guideline, recognition of care providers' intentions from their actions, and real-time as well as retrospective critiquing and quality assessment of care providers' application of he guideline, given the guideline and its intentions, the patient's medical record, and the analysis of the care provider's perceived intentions. In our previous work, we have designed a formal, machine-interpretable language, called Asbru, to represent and to annotate guidelines based on our framework, and implemented a small prototype interpreter for it. We will implement a full interpreter for the Asbru language and a set of software tools for specification, validation, and verification of Asbru guidelines. We will also design and implement a set of computational methods for runtime (during application) and retrospective quality assessment of clinical guideline application, both for individual patients and for populations of patients, using our intention-based model. We will rigorously test, validate, and evaluate our methodologies using collaborators in clinical areas such as oncology, hypertension, and endocrinology.