A bayesian network approach to disease subtype discovery

Abstract

Human diseases are historically categorized into groups based on the specific organ or tissue affected. Over the past two decades, advances in high-throughput genomic and proteomic technologies have generated substantial evidence demonstrating that many diseases are in fact markedly heterogeneous, comprising multiple clinically and molecularly distinct subtypes that simply share an anatomical location. Here, a Bayesian network analysis is applied to study comorbidity patterns that define disease subtypes in pediatric pulmonary hypertension. The analysis relearned established subtypes, thus validating the approach, and identified rare subtypes that are difficult to discern through clinical observations, providing impetus for deeper investigation of the disease subtypes that will enrich current disease classifications. Further advances linking disease subtypes to therapeutic response, disease outcomes, as well as the molecular profiles of individual subtypes will provide impetus for the development of more effective and targeted therapies.