HCLSIG/PharmaOntology/Meetings/2010-06-03 Conference Call

Conference Details

* Date of Call: Thursday June 3 2010 
* Time of Call: 12:00pm - 1:00pm ET 
* Dial-In #: +1.617.761.6200 (Cambridge, MA) 
* Dial-In #: +33.4.89.06.34.99 (Nice, France) 
* Dial-In #: +44.117.370.6152 (Bristol, UK) 
* Participant Access Code: 42572 ("HCLS2") 
* IRC Channel: irc.w3.org port 6665 channel #HCLS2 (see W3C IRC page for details, or see Web IRC) 
* Mibbit quick start: Click on mibbit for instant IRC access
* Duration: 1h 
* Convener: Susie

Agenda

Observational Medical Outcomes Partnership: Developing Tools for Conducting Observational Database Research Across a Network of Data Sources - Paul Stang
TMO Updates - Michel, Elgar
Patient Mapping - Michel, Chime, EricP
Interface/eMerge - Bosse, Chris
Sources of Patient Data - Scott
Outreach (IHI, UPenn Translational Medicine) - Michel, Susie
AOB

Minutes

Attendees: Chris, PAul, Bob, Chime, Joanne, Janos, Michel, Elgar, Matthias, BIn, Bosse, Susie

<Bob> Agenda: Paul Stang, updates of TMO patient data mapping, interface work by Chris

<Susie> http://esw.w3.org/images/5/5c/OMOP_presentation_ontology.pdf

<Bob> Paul: OMOP would be of interest for this group

<Bob> Paul: data are observational, US health financial records

<Bob> Paul: E games played in terms of coding, but does represent a totality

<Bob> Paul: EHR doc does not nec aggregate info

<Bob> Paul: non-randomized patients, healthcare as it is delivered!

<Bob> Paul: drug co: exploring safety carries reporting responsibilities

<Bob> Paul: slide 3 representation of flat file data

<Bob> Paul: claims DB, used for billing, to answer fairly simple questions

<Bob> Paul: slide 4 issue here is that prescriptions written may not be filled or used

<Bob> Paul: was patient actually exposed to med?

<Bob> Paul: she may be using inhaler upside-down, ergo ineffectively

<Bob> Paul: outcomes side E lots of issues w. codes

<Bob> Paul: other problem on outcomes, diagnoses peculiar in US

<Bob> Paul: may need incorrect "diagnosis" to get med, doctor needed code to get paid

<Bob> Paul: E missing data, like when someone dies

<Bob> Paul: slide 5 risks tend to be clinical, benefits tend to be quality of life

<Bob> Paul: but improvements may not be captured in data

<Bob> Paul: slide 6 it's all about the data, views are different!

<Bob> Paul: slide 7 OMOP is a true public-private partnership, trying to do experiments on what type of data and procedures

<Bob> Paul: can these data be employed to report earlier than present voluntary system?

<Bob> Paul: slide 8 governance, Paul is part of central research group

<Bob> Paul: looked for all possible methods

<Bob> Paul: OMOP Cup!! winner was means of spam filter, repurposed to this

<Bob> Paul: purchased 6 DBs, 5 DBs centrally, plus partners that move their own data to OMOP schems

<Bob> Paul: page 6 start methods, traditional plus refinements

<Bob> Paul: probably 1000 variations on methods to test

<Bob> Paul: next page, 1) E 10 drug-event pairs that we would expect to see

<Bob> Paul: 2) data mining to identify new associations

<Bob> Paul: TMO note ==> all of our work is freely from website

<Bob> Paul: including a simulated dataset of up to 100M people

<Bob> Paul: next month or so, will be refined data available

<Bob> Paul: we have a gold standard

<Bob> Paul: E extended consortium w peer review applied to common data model

<Bob> Paul: page 8 E software tools w. slightly different purposes

<Bob> Paul: E other automated ways of combining refills into one exposure block for many refills

<Bob> Paul: pictogram of map to common data model

<Bob> Paul: E white papers describing all of this, with all the cross-walks

<Bob> Paul: deployment of these methods across network is standarized

<Bob> Paul: next page Oscar and Nathan

<Bob> Paul: next page methods

<Bob> Paul: look at the OMOP website!

<Bob> Paul: in month 16 of 2-year project, getting preliminary outputs

<Bob> Susie: Q what are plans after month 24?

<Bob> Paul: going to pharma and funders about extensions, want to aggregate across multiple sites

<Bob> Bob: check out Netflix and blending

<Bob> Joanne: Q how can our 2 groups cooperate?

<Bob> Paul: E fear of what happens to data after Xformation?

<Bob> Paul: want to make that process easier and w. less pain

<Bob> Paul: getting data to better interop(?)

<Bob> Paul: E other efforts, our approach is empirical

<Bob> Paul: problem, people think that these data can solve all problems

<Bob> Joanne: how far along?

<Bob> Paul: 6 months before end, just getting results now

<Bob> Joanne: maybe a filter on data could help

<Bob> Susie: can we use ontologies to highlight obvious errors?

<Bob> Joanne: identifying those kind of problems, consistency checking

<Bob> Paul: everybody screaming for quality control, would have welcome at office of head coordinator

<Bob> Susie: modeling over to SAS?

<Bob> Paul: SAS, plus R, some in SQL sets

<Bob> Q on page 9 NDF = national drug file?

<Bob> Paul: yes, wanted to use ingredients, wanted class too

<Bob> Elgar: comment on criminal repurcussions

<Bob> Paul: DBs are not available

<Bob> Paul: if you are exploring from pharma then you have to follow regulatory rules!

<Bob> Paul: you are safe if you visit our website.

<Bob> Chime: digesting the framework..

<Bob> Susie: how can your CPR be aligned w OMOP?

<Bob> Chime: describe a little bit about alignment?

<Bob> Paul: it's just mappings of codes, some are standard, some are developed by GSK to map between vocabulary sets

<Bob> Paul: GSK(?) had multiple levels of review

<Bob> Paul: everything in data model was mapped

<Bob> Susie: interesting that OMOP using snomed, etc

<Bob> Susie: can this make it easier for us to connect to OMOP?

<Bob> Paul: yes, had connection to high-level people

<Bob> Joanne: how good are the simulated data?

<Bob> Paul: we had to pay for actual claims data, but we cannot act as intermediary

<Bob> Joanne: some data from CA, available?

<Bob> Paul: medicaid, medicare, etc. would dictate rules

<Bob> Paul: schools of pharmacy have become state guardians of data

<Bob> Paul: CMS have medicare data for whole country, claimed available for all US

<Bob> Paul: second set used monte carlo techniques, osim2 available in a month, better than first set

<Bob> Susie: Thank you Paul!

<Bob> Susie: think about potential for working w. OMOP

<Bob> Susie: is competiion a good openning for us?

<Bob> Susie: Next agenda: TMO update

<Bob> not yet

<Bob> Susie: Next agenda: Patient data mapping, want to make sure linked into other data sources

<Bob> Susie: Chime has CPR, Michel looked at BioTop?

<Bob> Chime: has notes shared with Michel

<Bob> Susie: maybe can have follow-up on Chime/Michel CPR

<Bob> Susie: next steps? mappings on these calls or separate calls?

<Bob> Chime: feedback on both CPR ont plus on mappings

<Bob> Chime: will put some issues on wiki, w. EricP

<Bob> Susie: wanting to use Indivo schema, but wan't ready in time

<Bob> Susie: since then Indivo is more broadly available

<Bob> Susie: should we move our patient data to Indivo?

<Bob> Janos: would be good(?)

<Bob> Susie: maybe Ken can join us

<Bob> Ben, not Ken!

<Bob> Chime: would be good to talk to Ben, also to give them feedback

<Bob> Janos: +1 very useful

<Bob> Both Ken and Ben!

<Bob> Susie: AOB? Chime can post general issues to wiki

<Bob> Susie: Chris eMerge, Marshfield

<Bob> Chris: is in process, nearish the end of this week, in touch w. Luke

<Bob> Susie: Paul's criminal repercussions, may be good for us to continue working w fake patient data!

<Bob> Chris: yes, Luke can answer whether this is working, ie w. fake data

<Bob> Susie: we do have good channels for adding fake data

<Bob> discussion of availability for real data

<Bob> need to keep things out of limelight because of this very bad regulatory issue

<Bob> get good authorization etc