Abstract
Robust biomedical ontologies contain the collective knowledge of the experts who created them, including elements of causality among symptoms. In addition to standardizing and integrating heterogenous symptoms, the structure and indexing within domain-specific biomedical ontologies can be sources of prior knowledge. This knowledge can be used to provide insight into disease progression and to infer causality, which is useful for orienting Causal Bayesian Networks (CBN) obtained from a variety of causal learning algorithms. This paper proposes a novel method of extracting and analyzing causal knowledge contained in two Authoritative Medical Ontologies (AMO), then testing the accuracy of the expertise. We will examine several ordered variable pairs from the ontological hierarchy in the Medical Dictionary for Regulatory Activities Terminology (MedDRA) and indexed terminology from the International Classification of Diseases Version 10 Clinical Modification (ICD-10-CM). These ordered pairs will be used to orient arcs in a learning algorithm to test their ability for improving conditional probabilities within a CBN based on patient data.