comprehensive ontology design for autism spectrum disorder
abstract
ontology is a formal explicit description of concepts in the knowledge domain. in recent
years, developing ontology in different domains is a hot topic for many researchers, especially in
the medical field because of the benefits offered to users. using ontology allows sharing and
reusing domain knowledge in an efficient and explicit way. in particular, ontology in medical
field can facilitate the access to query data, precise knowledge, and seamless sharing of
electronic medical records (emr). thus ontology increases the accuracy of doctor’s diagnostic
decision.
autism spectrum disorder (asd) often presents with difficulties in verbal and nonverbal
communication, behavior and social interactions. autism is difficult to define due to the
complex heterogeneous disorders in this domain and to the lack of coherent set of knowledge
that deals with all aspects of autism. the purpose of this research was to address these
shortcomings by developing a comprehensive asd ontology that formally conceptualizes both
domain and operational autism knowledge, unifies autism terminology, and facilitates access to
precise autistic information for both general public and expert users, thus enabling better
diagnostic and treatment decisions.
to build such ontology, we investigated many medical research works in the various
areas of autism such as disorders, effects and treatments of asd. the study was done with the
purpose of extracting and gathering information from the most trusted sources such as existing
ontologies, standard textbooks, relevant articles and clinical studies. these sources were used to
build a semantic map linking key concept classes. mainly we focused on properties and
relationships between these classes to formally describe the autistic domain and operational
knowledge and to bring the scattered knowledge into the ontological form. ontology
instantiation for each subclass was based on pilot studies and clinical cases.
the system was implemented using protégé, an ontological framework developed by the
stanford center for biomedical informatics research at the stanford university. the ontology
was built using the web ontology language (owl). owl is a semantic web language designed
to indicate the rich and complex knowledge of the domain. moreover, we developed a basic web
query system for the asd ontology to present the ontology information to different users around
the world. the developed system has been evaluated to measure quality of embedded
knowledge, ontology correctness and the usability of its web query system.