Managing uncertainty in diagnosis of acute coronaric ischemia.

Abstract

The paper presents the reasoning mechanism of COR, a knowledge-based system (KBS) able to provide support for the diagnosis of coronaric ischemia by integrating the interpretation of chest pain, 12-lead ECG, and bio-marker concentrations. Chest pain features are collected interactively through a questionnaire. The ECG signal is acquired in SCP format. Any set of bio-markers can be considered. Data input is incremental and possibly incomplete. Reasoning is based on revised uncertainty calculus, which allows a formal treatment of verbally expressed uncertainty concerning both input data and diagnostic rules. Each diagnosis is supplemented by a linguistic label, expressing the plausibility of the disease identified, given the symptoms observed.