Mining of health information from ontologies

Abstract

Data mining techniques can be used to efficiently analyze semi-structured data. Semi-structured data are predominantly used within the health domain as they enable meaningful representations of the health information. Tree mining algorithms can efficiently extract frequent substructures from semi-structured knowledge representations. In this paper, we demonstrate application of the tree mining algorithms on the health information. We illustrate this on an example of Human Disease Ontology (HDO) which represents information about diseases in 4 ?dimensions?: (1) disease types, (2) phenotype (observable characteristics of an organism) or symptoms (3) causes related to the disease, namely genetic causes, environmental causes or micro-organisms, and (4) treatments available for the disease. The extracted data patterns can provide useful information to help in disease prevention, and assist in delivery of effective and efficient health services