The use of ontology to represent nursing knowledge about family health history and facilitate automated search for clinical practice guidelines.

摘要

Representations of family health history commonly used in health care have two major limitations: they are limited to biological family relationships, and therefore cannot be used to represent the diversity of relationships found in real contemporary families; and as visual representations to aid human interpretation and analysis, they do not facilitate computer assistance for these cognitively challenging tasks. Use of such representations limits nurses' ability to represent and build knowledge about families. Nursing knowledge and practice may benefit from the use of computable representations that enable nurses to represent and build knowledge about real families, which increasingly include family members related by legal and emotional relationships in addition to biological ones.;The Family Health History Ontology (FHHO) was developed to enable representation of diverse family health histories and facilitate computer support for interpretation and analysis. Based on nursing conceptualization about family and family health history, the FHHO permits representation of diverse family relationships including biological, adoptive, foster, step, and fictive. Rules associated with the FHHO enable automated computation of biological relationships and family health history states from minimal data input.;Use of the FHHO to represent a dataset of 21 family health histories indicated that over 99% of family health history information could be represented, demonstrating complete domain coverage and resulting in a formalized knowledge base of 21 family health histories including 555 persons. Utility of the FHHO for nursing practice was demonstrated by an exercise in which the family health history criteria contained in 5 clinical practice guidelines were represented as rules. When executed, the rules automatically classified the 555 persons in the knowledge base as positive or negative for needing heightened screening regimens as recommended in the guidelines based on family health history. Kappa coefficients for interrater agreement between automated classification and manual classification by a domain expert ranged from 0.95 to 1.0, indicating near perfect agreement and success in automated classification. This successful evaluation indicated that the FHHO may be a useful information structure for applications supporting nursing practice and knowledge building. The FHHO is available from the National Center for Biomedical Ontology BioPortal at http://bioportal.bioontology.org/ontologies/38633.