Reference: Zegher-Geets, I. d. IDEFIX: Intelligent Summarization of a Time-Oriented Medical Database. 1987.
Abstract: The use of computerized medical records improves the quality of information but does not solve the problem of information overload. For this reason, it would be useful to have the ability to summarize automatically patients records into meaningful clinical events. IDEFIX is a knowledge-based system that produces intelligent summaries from a time-oriented database of patients who have systemic lupus erythematosus. Medical concepts in the system are represented by three entities of increasing complexity: abnormal primary attributes, abnormal states, and diseases. Abnormal states and diseases are derived from the abnormal primary attributes by IDEFIX using a combination of model and data-driven algorithms. Uncertainty associated with the derived states is handled with a Bayesian approach supplemented by Boolean predicates, using likelihood ratios. IDEFIX contains point-event predicates, which represent evidence about the temporal evolution of the disease. Likelihood ratios associated with point-event predicates are obtained from a transformation of the Internist-1 knowledge- base. Likelihood ratios combined with time-oriented predicates are computed through the use of time-oriented probabilistic functions. IDEFIX is coupled with a Display Module that, after summarization, generates interactive, graphical displays with optional explanation window.
Notes: 99 pages.