Re: Minutes from BioRDF/LODD telcon Monday 06/12/2012

On 06/14/2012 04:37 PM, Matthias Samwald wrote:
> I would also be interested in extending this with data extracted from
> European product labels. Compared to the US and Dailymed, this is far
> trickier. The data is not openly available for many countries, content has
> different languages, and structured product labels are even harder to come
> by.
This does sound very challenging. While in different languages, do you 
know if all 12 Euro countries follow the regulations regulated by 
Article 11 of Directive 2001/83/EC on the "Summary of Product 
Characteristics"? If so, it seems possible that the Article 30 (p32) of 
the that directive would be relevant (see attached). Note especially 
item #2:

"In order to promote harmonisation of authorisations for medicinal 
products authorised in the Community, Member States shall, each year, 
forward to the coordination group a list of medicinal products for which 
a *harmonised summary of product characteristics should be drawn up.*" 
(emphasis mine)

The "coordination group" here is the Committee for Medicinal Products 
for Human Use (CHMP) <http://tinyurl.com/7qcqqva>. Their webpage states:

"The CHMP publishes a European public assessment report (EPAR) for every 
centrally authorised medicine that is granted a marketing authorisation, 
setting out the scientific grounds for the Committee’s opinion in favour 
of granting the authorisation, plus a ‘summary of product 
characteristics’ (SPC), labelling and package leaflet (patient/user 
information leaflet) for the medicine, and details of the procedural 
steps taken during the assessment process. "

EPARs are available here: http://tinyurl.com/c8hua94

It looks like a whole bunch of harmonized SPCs (i.e., EU product labels, 
if I understand correctly) are available from this site. For example, I 
downloaded the attached SPC for Abilify (aripriprazole).

Hopefully, there is a standard structure to the document sections. 
Rimichi in [1] says that they are...

"Accordingly, SPCs of specialty medicines for human use are organized 
into 12 sections: name, therapeutic categories, active ingredient, 
excipients, indications, contraindication/side effects, undesired 
effects, posology, storage precautions, warnings, interactions, and use 
in case of pregnancy and nursing"

...but the one I looked at only has 10 sections. Any ways, it might be 
interesting to contact EPAR and see if you can get the documents used to 
create these SPC PDFs. Otherwise, one might wget all PDFs from the site, 
convert to text, and then fit the content into a SPL-like structure (in 
terms of sections) and provide as RDF.

-Rich

[1] Rubrichi S, Quaglini S. Summary of Product Characteristics content 
extraction for a safe drugs usage. J Biomed Inform. 2012 
Apr;45(2):231-9. Epub 2011 Nov 10. PubMed PMID: 22094356.
> A Viennese startup called http://www.diagnosia.com/de/ recently made package
> insert texts from 12 European countries openly available. I plan to contact
> them and see if they would also be interested in cooperating with research
> projects. We might not end up with a resource that can be shared without
> restrictions, but at least we could end up with something that allows us to
> run SPARQL queries for analysis and research.
>
> Other ideas for how such a multilingual, pan-European ressource are welcome!
>
> Cheers,
> Matthias
>
>
>> Hi, Richard,
>>
>> Very nice work. Building the linked data source for SPLs is very important
>> for facilitating its wide use. It is impressive that you keep the source
>> updated once per week.
>>
>> As you have pointed out that the SPLs are semi-structured, the combination
>> of NLP and standard terminologies will be crucial for producing structured
>> data out of the SPLs. As you know, our team at Mayo Clinic is
>> investigating
>> the SPLs from the perspective of Adverse Drug Events. We would like to
>> collaborate to see whether we could link our ADEpedia data
>> (http://adepedia.org) with your LinkedSPLs.
>>
>> The following are our preliminary works,
>>
>> 1)  Jiang G, Solbrig H, Chute CG. ADEpedia: a scalable and standardized
>> knowledge base of Adverse Drug Events using semantic web technology. AMIA
>> Annu Symp Proc. 2011;2011:607-16. Epub  2011 Oct 22.
>> http://www.ncbi.nlm.nih.gov/pubmed?term=adepedia
>>
>> 2) Zhu Q, Jiang G, Chute CG. Profiling Structured Product Labeling using
>> RxNorm and NDF-RT. ICBO 2012 Workshop (to be held in July 21-25, 2012).
>>
>> Best regards,
>>
>> Guoqian
>>
>>
>> Guoqian Jiang, M.D., Ph.D.
>> ===========================================
>> Assistant Professor of Medical Informatics
>> Division of Biomedical Statistics&  Informatics,
>> Mayo Clinic College of Medicine
>> 200 First Street, SW,
>> Rochester, MN, 55905
>> Tel: 1-507-266-1327
>> Fax: 1-507-284-0360
>> Email: jiang.guoqian@mayo.edu
>> ===========================================
>>
>>
>>
>>
>> On 6/14/12 12:47 PM, "Richard Boyce"<rdb20@pitt.edu>  wrote:
>>
>>> Minutes from BioRDF/LODD telcon Monday 06/12/2012
>>>
>>> Rich Boyce (University of Pittsburgh) presented "LinkedSPLs - a
>>> dynamically updated and comprehensive linked data node for US drug
>>> package inserts"
>>>
>>> Link to his slides:
>>> <http://www.slideshare.net/boycer/linkedsplsinitialhclspresentation06082012>
>>>
>>> Summary of discussion:
>>>
>>> Richard: Purpose of presenting is to attract collaborators...consider
>>> Linked Structured Product Labels an important data resource to keep up
>>> to date and improved, hopefully with partners.
>>>
>>> Richard: Structured Product Labels are the standard in the U.S.
>>> Medscape, Micromedex echo product labels and add information that is
>>> missing. If a recent IOM report (http://tinyurl.com/cxovshm) is acted
>>> on, we should see some progress in how the product labels are formed in
>>> the next few years
>>>
>>> ericP: q+ to ask if the SPLs reflect the areas of concensus and that
>>> affects the degree to which not being on the SPL indicates variation of
>>> information from the various resources
>>>
>>> Richard: (answering EricP): Interesting question. Warfarin is an example
>>> of how consensus from the community ended up in the product label. Often
>>> the information is not complete and it is left up to third parties to
>>> complete - dynamic process.
>>>
>>> Richard: FDA wrote laws the dictate the types of claims that should be
>>> present in each section. This gives us an opportunity to link to related
>>> resources such as for evidence.
>>>
>>> Richard: FDA wrote laws the dictate the types of claims that should be
>>> present in each section.
>>> ..This gives us an opportunity to link to related resources such as for
>>> evidence.
>>> ..[slide8]DailyMed is the public source of SPLs
>>> ..[slide9] "SPL IDs are not static, so a label's URL may change if
>>> ..the label is updated. However, DailyMed provides permanent
>>> ..URLs to view or download the latest version of an SPL²
>>>
>>> Richard: LinkedSPLs started with Anja's PHP for DailyMed and ported to
>>> python
>>>
>>> EricP: correcting for multiple active ingredients require modeling
>>> changes?
>>> Richard: Modeling changes
>>> UNII -
>>> Unique Ingredient Identifier
>>> sounds like Unicode when spoken!
>>> CODE: http://swat-4-med-safety.googlecode.com/svn/trunk/linkedSPLs/
>>>
>>> JoanneLuciano Health Care Delivery Grants from ???
>>>
>>> Simon Lee: Maybe use for AHRQ grants
>>>
>>> JoanneLuciano NIN NLM grants
>>> JoanneLuciano I'm thinking this can be linked with Safety-Code.org
>>>
>>> Richard: Think it's a little early for AHRQ grants, more research
>>> necessary
>>>
>>> Simon: agree.
>>> Simon: need working prototype.
>>> Simon Lin - Marshfield Clinic
>>>
>>> michel DrugBank is the best to link to, i think
>>> michel they then link out to many other sources
>>>
>>> JoanneLuciano safety-code.org
>>>
>>> michel BTW - I've built a new parser for drugbank
>>> michel
>>> http://bio2rdf.semanticscience.org:8022/describe/?url=http%3A%2F%2Fbio2rdf.org
>>> %2Fdrugbank%3ADB00975&sid=5
>>>
>>> michel we're in the process of adding this to bio2rdf
>>> mscottm asks michel - adding what?
>>> ericP zakim, please disconnect ericP
>>> Zakim EricP is being disconnected
>>>
>>>
>>> gamble @boycer I'm interested in data quality / Minimum information
>>> checklist validation potential for this data. Versioning is also
>>> interesting. I don't have a mic so I'll follow up by email.
>>> michel adding my parse of drugbank to bio2rdf
>>
>>
>>


-- 
Richard Boyce, PhD
Assistant Professor of Biomedical Informatics
Faculty, Geriatric Pharmaceutical Outcomes and Gero-Informatics Research and Training Program
Scholar, Comparative Effectiveness Research Program
University of Pittsburgh
rdb20@pitt.edu
412-648-9219 (W), 206-371-6186 (C)
Twitter: @bhaapgh