Abstract
Background: The American Board of Family Practice is developing a computer-based recertification process. An optimal implementation requires a formal model of family medicine, which will become the basis for a knowledge base.
Design: The proposed model of family medicine contains six entities: Population, Record, Agents of Change, Health States, Findings, and Courses of Action. The model illustrates 15 important relations between entities. For instance: Health States Lead to Health States, and Findings Associate with Health States. These two relations describe natural history, manifestations of disease, and the effects of medical interventions and risk factors. Because time is such an important aspect of primary care, nearly all numeric data are represented as graphs of possible values over time, called Patterns, which include details about periodicity. Patterns and other aspects of the model provide a means of describing covariance between observations, such as the influence of height on weight.
Results: The model reflects many family practice activities and suggests some formal descriptions of family practice. For instance, diagnostic activities focus largely on classifying early or short segments of Patterns in Findings. Most medical interventions attempt to alter either the probability distributions in a Lead-to relation or the impact of a Finding.
Conclusion: The proposed model of family medicine could find uses in many applications, including computer-based tests, medical records, reference systems, and decision support tools.