Health Assistance for Immigrants - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 85

Table 3.1 Concepts of the health ontology
Concept Description
Context This concept describes a composite information node composed of
property nodes and state nodes. A context concept for instance can describe
that certain therapies are only applicable for pregnant women
Property Is connected to a context node and specifies the context, for instance a
certain pregnancy week
Therapy/Procedure This node describes possible therapies to cure a disease
Cause Cause nodes describe the reason for a medical condition, e.g., obesity is a
cause for diabetes
ClinicalDiagnostics Describes an approach to make a diagnosis. For instance to take an x-ray of
somebody

Region Part of the human body plus psych. Indicates where a disease or injury is

situated

Prophylaxis This node describes prevention procedures, e.g., regular teeth-brushing

helps preventing cavities

Information The information node is the node that is displayed to the user with

information about the node it is connected too. This can be information
about diabetes, sport, pregnancy, or others. Information nodes can be a text
or website, a video or picture

State State is the general node to describe a condition of the body. Sub-nodes are

injury or disease

Injury Injury is a sub-node from State. It is splitted in inner and exterior injuries

InteriorInjury Internal Bleeding is an example for an InteriorInjury node

ExteriorInjury An abrasion is an example for exterior injuries

Symptom Symptom nodes define possible clinical signs that indicate a State

Disease Disease is a sub-node from State and is split into mental and physical

disease

MentalDisease This node comprises all diseases connected to a humans psych. For

example, depression is a MentalDisease

PhysicalDisease The PhysicalDisease node describes all body related diseases

Synonym The synonym node is important for all multilingual aspects of the described

system. It is explained later in Section Multilingual Semantic Search

as an example, the language-independent concepts Pregnancy and Malformation are
identified, since they have been labelled using the keyword “pregnant.” Furthermore,
the concepts Nutrition and Alcohol are identified by the label “besinleri.”
Using the retrieved query-relevant ontology concept nodes, we employ a graph-
search to find conceptually related information nodes. Different concepts from ontol-
ogy classes like diseases, diagnostics, and treatments are semantically linked with
weighted edges in our HO. Our algorithm performs a full graph search along these
edges, bounded in-depth. Information nodes found during this traversal are ranked
based on the proportion to the edge-weights of the path of the originating concept and
anti-proportional to that path's length. Information nodes found via multiple paths
receive the sum of the relevance values of all of those paths. Using the example query