Evaluation and Evolution of NAOnto – An Ontology for Personalized Diabetes Management for Native Americans

Ontology has been proposed to use in various healthcare domains to represent and manage knowledge, integrate heterogeneous data, assist natural language processing, facilitate machine learning, and support intelligent medical decisions. Although there are many ontologies defined, their adoption in real-world applications is limited, because their quality is unknown. The quality of the ontology directly determines its use and reuse in real-world applications, which is especially important for the medical domain. To assure its quality, an ontology should be evaluated with a set of comprehensive and systematic metrics and approaches. In this work, we studied the ontology evaluation problem in the medical domain. Specifically, we present the evaluation framework to evaluate, NAOnto, an ontology we developed for personalized diabetes management of Native Americans. The framework consists of two important parts of the evaluation, namely verification, and validation. We adopt software quality standard metrics to verify ontology as a software application while applying the ontology to real applications to validate its contained knowledge. Based on the evaluation results, the proposed ontology has gone through an active evolution process to address the discovered issues/problems. The proposed framework demonstrates how ontology evaluation can be practiced in the healthcare domain to assure the quality of ontologies.