Knowledge Engineering in Clinical Care

Abstract

During the care of critically ill patients, pertinent data may be obscured by the large amount of information in the patient record. we have developed a computer program to interpret measured physiological data, recognize the clinical context in which the data were measured and in which they must be interpreted, extract important data, and bring interpretations to the attention of the bedside clinician. The program uses a set of rules which contain “knowledge” about critical care patient management and “expectations” about the future possible interpretation of the patient physiology as it changes in time. The rules are statements of the expert clinical practitioner about the practice of critical care medicine. Each rule is of the form “if a set of facts are true,” then “make a conclusion based on these facts”. The rules are written in an English-like syntax using clinical terminology. Use of rules for representation and manipulation of knowledge about a problem is taken from the subdiscipline of Artificial Intelligence called “Knowledge Engineering”.