using semantic web technologies to reproduce a pharmacovigilance case study

12
Using Semantic Web Technologies to Reproduce a Pharmacovigilance Case Study Michiel Hildebrand, Rinke Hoekstra & Jacco van Ossenbruggen

Upload: jacco-van-ossenbruggen

Post on 05-Dec-2014

316 views

Category:

Technology


2 download

DESCRIPTION

We show how the use of PROV-O helps to analyse, discuss and communicate the reconstruction of a scientific workflow of a pharmacovigilance paper.

TRANSCRIPT

Page 1: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

Using Semantic Web Technologies to Reproduce

a Pharmacovigilance Case Study

Michiel Hildebrand, Rinke Hoekstra & Jacco van Ossenbruggen

Page 2: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study
Page 3: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

computational (open) data study

(open) data process results

prov:Entityprov:Entity prov:Entityprov:Entityprov:Activityprov:Activity

Page 4: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

pharmacovigilance

detect side effects of drugs: disproportional correlation between a drug and an associated adverse event

prov:Entityprov:Entity prov:Entityprov:Entityprov:Activityprov:Activity

Page 5: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

2x2 contingency table

Page 6: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

28.887

computation is never trivialcomputation is never trivial

28.663

28.767

28.862

28.86228.837

Page 7: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

?PROV helps to communicatePROV helps to communicate

Page 8: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

reproduction

2,231,038+9

1,664,078-142

?all drug names were unified into generic names by a text-mining approach. Spelling errors were detected by GNU Aspell and carefully confirmed by working pharmacists.

3.525+1

23,865,029+1,847,073

Foods beverages, treatments (e.g. X-ray radiation), and unspecified names (e.g. beta-blockers) were omitted

debugging requires intermediate datasetsdebugging requires intermediate datasets

Page 10: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

reproduction

original

PRR = 2.520

PRR = 2.504

Page 11: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study
Page 12: Using Semantic Web Technologies  to Reproduce  a Pharmacovigilance Case Study

PROV helps to communicate

>> share your provenance graph

PROV helps to communicate

>> share your provenance graph

debugging requires intermediate datasets

>> share each prov:Entity

debugging requires intermediate datasets

>> share each prov:Entity

computation is never trivial(applies also to “preprocessing” & “well known” formula’s)

>> share each computational prov:Activity

computation is never trivial(applies also to “preprocessing” & “well known” formula’s)

>> share each computational prov:Activity