Cluster Methodology Defines Archetype Sentinel Consomic Rats

A clustering method was developed to identify rats demonstrating characteristics of hypertension (\´sentinel\´ animals). Although all rats in the group do not demonstrate every symptom of hypertension, these \´archetype sentinels\´ do demonstrate essential characteristics of the disease. This method applies two computational techniques to create a mechanism for classifying individuals whose biomedical profile include a spectrum of disease related phenotypes. First, fuzzy cluster means (FCM) is used to distinguish a very small group of archetype subjects from the general population. Archetypes are defined as those subjects that are most often classified correctly within a limited portion of the entire collection of biomedical phenotypes. The resulting archetype is then examined to determine which phenotypes best characterize the set. In this study, the defining set of phenotypes demonstrate essential symptoms of hypertension. Consequently, we consider the archetype set \´sentinel\´ animals for hypertension. The archetype sentinels are then used to train a neural network (NN) to determine the physiological characteristics of pheno-typed subjects which are not the archetype sentinels. Two inbred strains of rats are used: Brown Norway (BN) and the Brown Norway/Salt Sensitive Chromosome 20 (SS20BN). The physiological database contains a total of 63 phenotypes (41 renal, 22 cardiac). A total of 79 phenotyped rats were analyzed with the FCM method yielding 6 BN archetypes and 5 SS20BN archetype subjects characterized by a total of 39 phenotypes (18 renal and 21 cardiac.) These 11 archetype sentinels then were used as a neural network training set to classify the non-archetype sentinel subjects as either normal (BN) or hypertensive (SS20BN). Of all rats tested, 10 of 11 BN and 10 of 10 SS20BN rats were properly classified. Overall, 95% of the rats were classified correctly, with one false positive result. Results demonstrate that the FCM method can be used to isolate the "- rchetype Sentinels." These archetype sentinels can then be used to train a perceptron neural network to determine classification of unrelated rats with the same genetic background. This approach can be generalized and used to classify other disease and normal rat models. In addition, the method may be applied to human population studies in a similar manner