There is a new technology to express and search the data that can provide more meaning and relationship –
semantic technology. The semantic technology can easily add, change and implement the meaning and relationship
to the current data. Companies such as Facebook and Google are currently using the semantic technology. For
example, Facebook Graph Search use semantic technology to enhance more meaningful search for users.
The paper will introduce the basic concepts of semantic technology and its graph data model, Resource Description
Framework (RDF). RDF can link data elements in a self-describing way with elements and property: subject,
predicate and object. The paper will introduce the application and examples of RDF elements. The paper will also
introduce three different representation of RDF: RDF/XML representation, turtle representation and N-triple
representation.
The paper will also introduce “CDISC standards RDF representation, Reference and Review Guide” published by
CDISC and PhUSE CSS. The paper will discuss RDF representation, reference and review guide and show how
CDISC standards are represented and displayed in RDF format.
The paper will also introduce Simple Protocol RDF Query Language (SPARQL) that can retrieve and manipulate data
in RDF format. The paper will show how programmers can use SPARQL to re-represent RDF format of CDISC
standards metadata into structured tabular format.
Finally, paper will discuss the benefits and futures of semantic technology. The paper will also discuss what semantic
technology means to SAS programmers and how programmers take an advantage of this new technology.
2. Agenda
➢Introduction of Semantic Technology
➢Introduction of RDF & SPARQL
➢Applications in Clinical Data Life Cycle
➢Final Thoughts
➢Questions and Discussion
3. Introduction of Semantic Technology
➢What is Semantic Technology?
➢A new way to observe and model data
5. Data Model of Semantic Technology
➢ Basic concepts
➢ Consistent meaning of data
➢ Relationship between data
➢ Basic data model: a triple
➢ Subject
➢ Predicate / relationship / property
➢ Object
subject object
predicate
7. Example 2
➢“Kevin lives in Philadelphia” and “Kevin is an SAS
programmer” and “Kevin attends PharmaSUG”
Kevin Philadelphia
livesIn
SAS
programmer
is
PharmaSUG
attend
11. Facebook Graph Search
A semantic search
engine that was
introduced by Facebook
Question: Restaurants
that Kevin’s friend likes in
Philadelphia
12. Facebook Graph Search
Kevin Helen
Penn’s
Landing
PhiladelphiaRestaurant
isaFriend like
type isLocatedIn
Restaurants that my friends like in Philadelphia.
13. Semantic driven Questions in
HealthCare?
➢Find the patients who have the same
symptoms that I have and who take the same
drugs?
➢Find the patients who have the same AE as I
did while using drug 1.
14. Introduction of RDF
➢ A standard data representation of Semantic Technology
maintained by www.w3c.org
➢ Data structures
➢ RDF graphs: sets of subject-predicate-object
<http://rdf.cdisc.org/std/sdtmig-3-1-3#Column.DM.AGE>
mms:dataElementLabel
"Age"^^xsd:string .
➢ RDF datasets: collection of RDF graphs (e.g., sdtmig-3-1-3.ttl)
DM:AGE “Age”
DataElementLabel
15. Normal Table vs RDF Data
SUBJID SITEID SEX AGE AGEU RACE
001 01 M 45 YEARS WHITE
002 01 F 38 YEARS ASIAN
DM:001 001subjid
01
site id
M
sex
45
age
YEARS
ageu
WHTIE
race
16. Introduction of CDISC RDF
➢ CDISC Standards in RDF
➢ CDISC Standards in RDF Reviewer guide
➢ CDISC Standards in RDF Reference guide
➢ RDF representation in
https://github.com/phuse-org/rdf.cdisc.org
17. Introduction of SPARQL
Simple Protocol RDF Language (SPARQL) – a
standard query language that can convert
RDF graph format to structured format
18. Examples of SPARQL
Q: What is description of DM.AGE in sdtmig-3-1-3.ttl?
prefix mms: <http://rdf.cdisc.org/mms#>
select ?o
where { <http://rdf.cdisc.org/std/sdtmig-3-1-3#Column.DM.AGE>
mms:dataElementDescription ?o }
"Age expressed in AGEU. May be derived from RFSTDTC and
BRTHDTC, but BRTHDTC may not be available in all
cases (due to subject privacy concerns)."
19. PhUSE Semantic Technology
Working Group
➢Analysis results and metadata in RDF – develops
standards models and technical standards for the storage
and usage of analysis results data and metadata using
RDF data cube and R package.
➢Clinical program design in RDF – develops a RDF model to
capture, retain, reuse and share the design of clinical
programs.
➢Regulations in RDF – develops a searchable resource by
extracting and linking structured information from
regulations, guidance and regulatory processes.
➢Use cases for linked data – develops use cases for linked
data solutions in clinical data life cycle.
20. End to End Clinical Trial Linked
Artifacts Development
Protocol
Cheson
2007
Collectio
n
Tumor
Measuremen
t
SDTM
TR
Analysis
Progression
Free
Survival
Time to
Even
Analysis
Report
ADaM
ADTTEPFS
Bone Marrow
Assessment
Spleen and
Liver
Enlargement
FA
TU
LB
Response
PE
RS
23. Real Word Data Integration with Clinical
Trial Data
Clinical Trial Data
Colon
cancer
46
Male
Clinical Study
001
Tumor
Lesion
image
20 mm
15 mm
Drug1
Sex
symptom
age
Year
unit
participate
findings
drug
method
CT
4
0
week
Patient 1
me
result
result
weekSex
age
symptom
24. Final Thoughts
➢New technology
➢ Emerging technology
➢ Here to stay.
➢Contextual meaning
➢Connects to unlimited data.
➢ Links to public linked data (e.g., DBPedia and OpenData)