PREDOSE: PRE D O-S E

PREDOSE:

PRE
SCRIPTION

D
RUG

ABUSE

O
NLINE
-
S
URVEILLANCE

AND

E
PIDEMIOLOGY


Delroy Cameron

9/29/2011


GOAL


To determine user
knowledge
,
attitudes
and
behavior

related to the non
-
medical use of
pharmaceutical
opioids

as discussed on Web
-
based forums



To determine
temporal trends

and
patterns
in
pharmaceutical
opioid

abuse as discussed on
Web
-
based forums


2

P
ROBLEM



Scalability


Data Collection



Flexibility


Data Annotation



Latency of Relevant Information


Reactive Effect

3

A
SPECTS


Information Extraction


Entity Identification/Disambiguation


Relationship Extraction


Triple Extraction



Trend Detection


Spatio
-
Temporal
-
Thematic Analysis


Sentiment Analysis

4

Web
Crawler

Informal Text Database

Web Forums











Machine
Learning

Natural
Language
Processing

2

4

8

Data Cleaning

Stage 1. Data Collection

3

Stage 2. Automatic Coding

Stage 3.
Data Analysis and Interpretation

1

7

Qualitative and Quantitative Analysis
of Drug User Knowledge, Attitudes
and Behaviors

Named Entity Identification,
Relationship Extraction

+

=

Semantic Web Database

Information Extraction Module

Temporal Analysis for Trend Detection

Cuebee

Semantic Web Tools

9

10

Ontology

Triples/RDF Database

Scooner





















5

6

Schema

Instances

e.g. Opioid,
Pain Pills

e.g. Suboxone,
Subutex

E
NTITY

D
ISAMBIGUATION








Semantic Similarity

oxy

Oxycontin

OP

Oxycontin


Oxycontin

OC

Oxycontin

OC

oxy

“Experiment

2
:

Fail

-

Grinding

up

and

parachuting


-

despite

milling

these

OPs

down,

they

still

retain

substantial

time

release
.

I

found

this

to

be

a

failure

and

it

released

the

oxy

slowly

over

the

course

of

many

hours
.

oxy

Oxycontin

OP

oxy

Oxycontin


bad boys

Oxycontin

OP

Context

Candidates

6

P
ROBABILISTIC

E
NTITY

D
ISAMBIGUATION


“Experiment 2: Fail
-

Grinding up and parachuting

-

despite milling
these
OPs

down, they still retain substantial time release. I found this
to be a failure and it released the oxy slowly over the course of many
hours

.”






7

Vocabulary

Known Slang/Drug References

Informal Text
Database

Language
Model

NIDA

Erowid

NDCP

DrugSlang

Doc

Bluelight.ru

Opiophile.com

DrugsandBooze.com

Drugbank

Vocabulary

Corpus/Subset

Entity Identification/Disambiguation Architecture

8

S
ENTIMENT

A
NALYSIS






Entities


Oxy, OP



Sentiment Clues


failure, despite



Polarity

“Experiment

2
:

Fail

-

Grinding

up

and

parachuting


-

despite

milling

these

OPs

down,

they

still

retain

substantial

time

release
.

I

found

this

to

be

a

failure

and

it

released

the

oxy

slowly

over

the

course

of

many

hours
.

9

T
EMPORAL

A
NALYSIS

(TIMELIME)

0
100
200
300
400
500
600
700
800
900
2001-09
2003-06
2004-07
2004-11
2005-03
2005-07
2005-11
2006-03
2006-07
2006-11
2007-03
2007-07
2007-11
2008-03
2008-07
2008-11
2009-03
2009-07
2009-11
2010-03
2010-07
2010-11
2011-03
2011-07
Oxycontin (All)

0
100
200
300
400
500
600
2001-09
2003-06
2004-12
2005-04
2005-09
2006-01
2006-05
2006-09
2007-01
2007-05
2007-09
2008-01
2008-05
2008-09
2009-01
2009-05
2009-09
2010-01
2010-05
2010-09
2011-01
2011-05
2011-09
Oxycontin (Bluelight)

0
100
200
300
400
500
600
2005-02
2005-08
2006-01
2006-05
2006-09
2007-01
2007-05
2007-09
2008-01
2008-05
2008-09
2009-01
2009-05
2009-09
2010-01
2010-05
2010-09
2011-01
2011-05
2011-09
Oxycontin (Opiophile)

0
50
100
150
200
250
300
2004-03
2004-10
2005-04
2005-08
2005-12
2006-04
2006-08
2006-12
2007-04
2007-08
2007-12
2008-04
2008-08
2008-12
2009-04
2009-08
2009-12
2010-04
2010-08
2010-12
2011-04
2011-08
Oxycontin (DNB)

10

P
EOPLE



kno.e.sis


Delroy Cameron


Sujan

Udayanga


Amit

P.
Sheth



CITAR
(Center for Interventions, Treatment and
Addictions Research)


Dr. Raminta
Daniulaityte


Dr. Robert Carlson


Russel

Falck

11




Questions

http://wiki.knoesis.org/index.php/PREDOSE


12