DIAGNOSING CONGENITAL HEART-DEFECTS USING THE FALLOT COMPUTATIONAL MODEL

This paper describes a computational model developed for the diagnosis of multiple defects. If multiple defects interact, meaning that the c ues observable for multiple defects are not a sum of the cues observab le for the component defects, diagnosis is particularly difficult. We developed a description and classification of the ways cues change whe n defects interact. A computational model (named Fallot) was implement ed and a knowledge-base was constructed for the diagnosis of congenita l heart defects. On each case, Fallot performs recognition-based reaso ning followed by solution construction and evaluation with the cue com bination methods. Fallot was tested on cases from hospital files and c orrectly diagnoses cases with multiple interacting defects for which c onventional methods are not applicable or fail. (C) 1997 Elsevier Scie nce B.V.