Latent variable models for statistical diagnosis

Abstract

SUMMARY A class of statistical models is derived for the problem of differential diagnosis between several types of disease. The performance of models for differential diagnosis between three and four types is assessed on clinical data.