Research Abstracts - 2006

Knowledge Representation in Medicine: Electronic Clinical Guidelines

John Perry

In 2004, the Rand Corp reported that patients receive evidence-based medicine only 55 percent of the time and potentially harmful medicine 11 percent of the time [1]. The electronic clinical guideline (in contrast to the widely used paper-based clinical guideline) is just one of computation's contributions to improving these statistics. Electronic clinical guidelines aim to improve the quality and consistency of care by reducing inappropriate variations in practice and promoting the delivery of evidence-based medicine. It is critical that a guideline's knowledge representation be expressive, make ontological commitments, allow reasoning, and be efficient [2]. We have surveyed the electronic guideline literature, as well as the workflow literature, to identify fundamental components needed to design a proper guideline formalism [3]. In addition, we have researched a significant component of guidelines: time series matching, a.k.a. case-based reasoning, which includes model abstractions, feature extraction and reduction methods, and multidimensional indexing methods.

Funded by the National Library of Medicine

References:

[1] RAND Corp. The First National Report Card on Quality of Health Care in America. 2004.

[2] Randall Davis and Howard E. Shrobe and Peter Szolovits. What Is a Knowledge Representation? AI Magazine, 14(1):17-33, 1993.

[3] Samson W. Tu and Mark A. Musen. Representation Formalisms and Computational Methods for Modeling Guideline-Based Patient Care. 2000.