Health Assistance for Immigrants - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 86

Table 3.2 Relations of the health ontology
Relation Description
isTheReasonFor Connects Cause and State nodes. For example, obesity is caused by bad eating
habits
determines Connects ClinicalDiagnostics and State. Application: Obesity is determined by
measuring the BMI
Affects The affects edge defines effects that one State has on anther. For example,
Obesity affects Diabetes
Prevents The prevents edge connects the Prophylaxis and State node. A healthy diet
prevents obesity
refersTo refersTo connects Symptom nodes and State nodes, e.g., breathing problems
indicate obesity
treats The influences edge connects the context node with the Region, Prophylaxis,
Therapy/Procedure, Cause and ClinicalDiagnostics node
influences This node describes prevention procedures, e.g., regular teeth-brushing helps
preventing cavities
isAffectedBy Connects Region node with State node, such as Psyche is affected by
depression
Describes The describes edge connects information with the Property, Context,
Prophylaxis, Therapy, State node. Thereby, we can add information (pdfs,
websites, etc.) to a node
influencesCause Connects Therapy/Procedure with Cause nodes. E.g., the treatment of
respiratory distress is dependent on whether the cause is asthma or bronchitis

Has has connects Context and Property nodes. This allows defining a special

context. For example, the state pregnant can be constrained with the Property
9th month, to indicate that a treatment is only allowed in the 9th month

isRelatedTo Connects context with state. See example above

Symptom Symptom nodes define possible clinical signs that indicate a State

notationOf Connects Synonyms with all type of nodes to add multilingual information

above, we find among others an information node concerning Alcohol Consumption
during Pregnancy . The information node is ranked highest because it is linked very
closely to all of the three health concepts found for our keywords. As mentioned
above, information nodes in the HO have documents attached to them, which are
used as search results. These documents can be of any language, and have a lan-
guage tag attached. Direct translations of documents are marked as copies. In order
to gather search results for visualization, we collect the documents attached to the
most relevant found information nodes during the graph search step and rank the list
of documents depending on their relevance values combined with the user's language
preferences. Depending on these values, it is possible for users to find documents
in a different language near the top of the list, if the document is not available in
the preferred language but is highly relevant. Thus, users will always see a list of
results balanced by relevance and their language preferences. The results on this list
are independent of the language of the search query, since the results were found via
a mapping to language-independent ontology concepts.