M Schnabel1. 1. Institute of Medical Statistics and Epidemiology, Technical University, Munich, Germany. manfred.schnabel@imse.med.tu-muenchen.de
Abstract
OBJECTIVES: The aim is to show the flexibility, adequateness, and generality of formal concept analysis (FCA) applied to expert systems in medicine. METHODS: The basic idea of formal concept analysis is to look at a set of objects together with their attributes (formal context) under a definite mathematical view. This view leads to a mathematical structure, a complete lattice, which can be represented graphically. RESULTS: Some examples show that this method is very general and can be used to describe diseases, relationships between diseases and findings, the inference process, and among others, types of uncertainty. For many applications, the adequateness of this method, concerning the underlying semantics, can easily be made plausible. CONCLUSIONS: FCA can be used to analyze data that can be described by objects and attributes of any kind. The selected examples (diseases, patient cases, therapeutic decisions, rules) show the usefulness of this method. Although it is not difficult to transform the relevant semantics into a formal context in many cases, much more experience is necessary.
OBJECTIVES: The aim is to show the flexibility, adequateness, and generality of formal concept analysis (FCA) applied to expert systems in medicine. METHODS: The basic idea of formal concept analysis is to look at a set of objects together with their attributes (formal context) under a definite mathematical view. This view leads to a mathematical structure, a complete lattice, which can be represented graphically. RESULTS: Some examples show that this method is very general and can be used to describe diseases, relationships between diseases and findings, the inference process, and among others, types of uncertainty. For many applications, the adequateness of this method, concerning the underlying semantics, can easily be made plausible. CONCLUSIONS: FCA can be used to analyze data that can be described by objects and attributes of any kind. The selected examples (diseases, patient cases, therapeutic decisions, rules) show the usefulness of this method. Although it is not difficult to transform the relevant semantics into a formal context in many cases, much more experience is necessary.