Abstract

A combinatorial method is described for grouping cases with multiple malformations for the purpose of identifying previously undescribed syndromes. This method includes ways of carrying out 'tight' or 'loose' grouping, of allowing for variability of reporting of physical features by different observers, and of minimising the number of 'spurious' groups. Evaluation using a test data set of known dysmorphic syndromes showed that the method provides a feasible and useful means of grouping undiagnosed cases.