BROWSE
Detail
Conference Paper
Stroke Medical Ontology for Supporting AI-based Stroke Prediction System using Bio-Signals
Cited 7 time in 
Share 
Share
- Authors
- SoonHyun Kwon, Se Jin Park, Jaehak Yu, Jong-arm Jun, Cheol Sig Pyo
- Issue Date
- 2021-08
- Citation
- International Conference on Ubiquitous and Future Networks (ICUFN) 2021, pp.53-59
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- In this paper, we propose a stroke medical ontology that provides medical knowledge to accompany AI-based stroke disease prediction system's results that were arrived at based on EMG information. This system was developed as a result of the limitations mentioned above being encountered in previous studies. We approached the problem from a viewpoint of knowledge engineering with the aim of modeling medical knowledge related to strokes. Using web ontology language (OWL), a standard ontology language, we developed schema-level stroke ontologies with concepts and properties based on the brain's anatomical structures, lesions, and disease related to strokes. Also, we developed an instance-level medical terms ontology that can span standard medical terms such as those in the international classification diseases (ICD), systematized nomenclature of medicine - clinical terms (SNOMED-CT), and foundational model of anatomy (FMA). The above schema ontology and instance ontology are meaningfully mapped to each other to apply layered ontology modeling techniques that separate schemas from instances. Through semantic web rule language (SWRL)-based inference, we predict lesions, diseases, and anatomical brain structural ripple effects based on the patient's current lesions and diseases. The inferred knowledge information is provided via the SPARQL protocol and RDF query language (SPARQL), a standard ontology query language. To verify the stroke medical ontology proposed in this paper, we developed an ontology-based stroke disease prediction system. This system achieved knowledge augmentation performance of 67.82% by comparing the patients' current lesions and diseases with the lesions, diseases, and areas of disability found by SWRL-based inference using actual stroke emergency data from 37 patients.
- KSP Keywords
- Disease prediction, Knowledge augmentation, Knowledge engineering, Medical Knowledge, Medical ontology, Modeling techniques, Ontology query, Prediction System, Ripple effects, SNOMED-CT, Web ontology language