Data	harmonization	of	
Medical	Records	for	
federated	search
Radovan	Tomášik
BBMRI.cz
Structure
15min	presentation
Context
Harmonization
5min	for	Q&A
BBMRI.cz
Introduce	czech	biobanks	to	the	21st	century
Part	of	the	BBMRI-ERIC
Provide	support	and	infrastructure
BBMRI-ERIC
Eouropean	reseach	infrastructure	
for	biobanking
Provide	tools	and	software	
solutions
Problems	with	sharing	Medical	data
Hoarding opposed	to	Sharing
Convenience	opposed	to	Standardization
Conservatism	opposed	to	Innovation
Sharing	samples
Publicly	list	
samples	available	
for	research
What	other	data	
sets	are	available	
for	those	samples	
Centralized	vs	Federated	Search
Centralized	search
Biobank	n
Central	
Storage Biobank	1
Biobank	2
Federated	Search
Central	component
Local	components	of	federated	search
Install	local	
components	behind	
the	institution's	
firewall
01
Open	specific	ports
02
Limit	access	from	
within	the	
instituition
03
Extract	Transform	Load
Extract
Create	export	
from	(usually	2)	
databases
Transform
Harmonize	the	
data	into	a	
standard	format	(	
OMOP,	FHIR)
Load
Upload	the	data	
into	the	local	
component
Transform/Harmonize
XML,	CSV,	
JSON
Magic
FHIR.json
OMOP.csv
Example	OMOP	File
OMOP	CDM
The	Observational	Medical	Outcomes	Partnership	Common	Data	
Model
FHIR
Fast	Healthcare	
Interoperability	Resources
Framework	comprised	of	92	
so	called	“Resources”
Usable	for	a	variety	of	
medical	data
Example	
XML	patient	
file
Example	
FHIR	File
Issues	that	
need	to	be	
addressed
Mapping
Extending	the	data	model
Updating	available	samples
Updating	patient	consent
Correct	formats
Mapping
Impossible	to	do	a	
general	mapping	guide
1:1,	1:2	or	even	1:3	
mapping
Biobanks	don’t	have	the	
IT	resources	needed	
Leave	the	biobanks	to	do	the	
mapping	themselves
• Expensive	for	the	biobanks
• hard	to	maintain
• No	benefits	for	others
• Off-putting
• Takes	too	long
Support	the	biobanks	by	providing	
a	set	of	tools	for	setting	up	the	ETL	
pipeline	
• Documentation!
• GUI	app	for	less	tech	savvy	
biobanks
• Libraries	for	more	demanding	
cases
Questions?
Thank	you	for	your	
attention
Radovan	Tomášik