Knowledge-based ECG interpretation: a critical review

This work presents a brief review of some selected knowledge-based approach es to electrocardiographic (ECG) pattern interpretation for diagnosing vari ous malfunctions of the human heart. The knowledge-based approaches discuss ed here include modeling an ECG pattern through an AND/OR graph, a rule-bas ed approach and a procedural semantic network (PSN) based approach for ECG interpretation. However, certain syntactic approaches to ECG interpretation are also covered, considering their precursory roles to knowledge-based EC G interpretation. A fuzzy-logic-based approach is included in the discussio n to show how imprecision can be dealt with in modeling cardiological knowl edge. A domain-dependent control algorithm is discussed to show how the pro duction level parallelism can be exploited to reduce the length of the matc h-resolve-act cycle of a rule based ECG interpretation system. The review a lso contains a brief description of some recent applications of connectioni st approaches to ECG interpretation. This discussion finally ends with a co mparative assessment of performances of all the above-mentioned knowledge-b ased approaches to ECG interpretation and some hints about the future direc tions of work in this held. (C) 2000 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.