HCLSIG/PharmaOntology/Meetings/2010-05-27 Conference Call

From W3C Wiki

Conference Details

* Date of Call: Thursday May 27 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

  • CPR Presentation - Chime
  • TMO Updates - Michel, Elgar
  • Patient Mapping - Michel, Chime, EricP
  • Sources of Patient Data - Scott
  • Provenance Requirements - Joanne
  • Interface/eMerge - Bosse, Chris
  • Outreach (IHI, UPenn Translational Medicine) - Michel, Susie
  • AOB


Minutes

Attendees: Chris, Elgar, Chime, Matthias, Janos, Bob, Michel, Bosse, Scott, Christi, Joanne, Susie

  • <matthias_samwald> scribenick: Bob
  • <Bob> Chime: why not re-use Galen or Snomed, galen old, snomed has issues
  • <Bob> Chime: bfo; then biotop meant to integrate medicine and life sciences
  • <Bob> Chime: bfo well organized, biotop gets us to life sci, snomed, most recently Continuity of care CCR
  • <Bob> Chime: biotop framework acrross granularities; proposed as upper ont for snomed
  • <Bob> Chime: biotop GO, KEGG; has some peculiarities, but goes from top to places for sequences, etc
  • <Bob> Chime: E some alignment w FMA, and between FMA and snomed
  • <Bob> Chime: cpr adheres to realist phil; Rector on how to capture distinctions
  • <Bob> Chime: cpr captures clincial care model
  • <Bob> Chime: causes of disease; E ont of medical reality
  • <Bob> Chime: rule of thumb: distinctions are useful for real-world problems
  • <Bob> Chime: tried to validate patient data against terms and ont
  • <Bob> Chime: 'clinical finding' is key class
  • <Bob> Chime: clinical finding is outcome of diagnostic activity
  • <Bob> Chime: disease model, manifestations
  • <Bob> Chime: disease usually localized to anatomy
  • <Bob> Chime: symptoms are subjective observations by patient
  • <Bob> Chime: sign is result of observation by clinician
  • <Bob> Chime: clinical diagnosis is artifact recorded in patient record
  • <Bob> Chime: have 200K patient RDF dataseet
  • <Bob> Chime: cpr validated against snomed and fma; took advantage of where snomed fits in bfo framework
  • <Bob> Chime: E opportunity to validate against TMO patient data
  • <Bob> Susie: should be integrate tmo w. cpr
  • <Bob> Chime: we can align either bottom-up or top-down
  • <Bob> Chime: maybe we can look moving away from biotop (might not have this scribed right)
  • <Bob> Susie: should look again at biotop and how we fit
  • <Bob> Scott: sign implies that someone has disease
  • <Bob> Scott: does sign determine that E disease?
  • <Bob> Chime: sign is hypothetical, unclear how to capture strong relationship
  • <Bob> Chime: sign/disease relation is anecdotal, not part of clinical reasoning process
  • <Bob> Chime: would rely on physician to make diagnosis, and whether E diagnosis in record
  • <Bob> Chime: definitely a problem, esp w. clinicians
  • <Bob> Chime: want to capture how symptom or sign is captured
  • <Bob> Michel: object to sign/symptom from realist perspective
  • <Bob> Michel: we must come up w framework, but might not jive w. other formalism
  • <Bob> Michel: but want to make sure useful in query of patient records
  • <Bob> Janos: need translation to get to snomed?
  • <mscottm> Scott (before Michel): Is it possible to use signs and symptoms such that they are recorded as observations (and the implication of disease can be found but is not 'hard coded' into the ontology)? i.e. hypothetical knowledge is still captured in crisp logic
  • <Bob> Chime: had to determine what our terms were, then place them in cpr
  • <Bob> Q: you have access to Cleveland dIDed data?
  • <Bob> Chime: this is significant challenge; every time we need data we need to remove dates etc.
  • <Bob> Chime: we anonymize data; but on my own I can find whether I can align w. data that I have
  • <Bob> Susie: interested in validating cpr against cpr data?
  • <Bob> Susie: are you prepared to change cpr?
  • <Bob> Chime: yes, very motivated to map to tmo
  • <Bob> Chime: cpr is not cast in stone, looking to make sure schema useful
  • <Bob> Chime: want to ontologize w. use cases and w. patient data, even if synthetic
  • <Bob> Michel: we've got use cases, been working with Chime
  • <Bob> Michel: pragmatic approach will ensure success
  • <Bob> Q can we use extraction from fma and snomed
  • <Bob> Chime: all our tools are open source, so yes
  • <Bob> Chime: requires access to snomed and fma
  • <Bob> Chime: our famework can take a list of snomed terms and fma terms and map them together
  • <Bob> Q using homebaked URIs?
  • <Bob> Chime: most recent snomed has a URI scheme
  • <Bob> Susie: Michel Elgar updates on tmo itself?
  • <Bob> Elgar: next few days
  • <Bob> Susie: encouraging people to get back with comments on TMO
  • <Bob> Michel: looking at cpr and ogms
  • <Bob> Susie: Patient Mapping agenda
  • <Bob> Michel: pretty sure can add mapping to ogms and cpr, maybe we can extend?
  • <Bob> Susie: should be take another look at biotop?
  • <Bob> Michel: biotop has huge # of categories, but don't know how meaningful
  • <Bob> Michel: we have basic classes and basic roles
  • <Bob> Michel: questionable how useful so many classes etc
  • <Bob> Susie: having a patient-focussed approach seems to make a lot of sense
  • <Bob> Susie: Sources of patient data
  • <Bob> Scott: will make a wiki page
  • <Bob> Scott: Wendy Chapman says to sign up for access to NIH?
  • <Bob> Wendy has several thousand patient reports
  • <Bob> Plus ODIE from Pittsburgh
  • <Bob> Scott: these might be good for Chime too
  • <Bob> Scott: 1) Wendy's source 2) Rebecca Crowley has a set of pathology reports, but need to be associated with instituions
  • <Bob> discussion of LODD
  • <Bob> provenance task force has a knowledge base
  • <Bob> Scott: will try to post on wiki page
  • <Bob> Susie: Provenance from Joanne
  • <Bob> Joanne: Elgar and I have met off-line
  • <Bob> Joanne: will take time and effort to understand this issue
  • <Bob> Joanne: Workshop coming up on provenance; will come back w. better understanding
  • <Bob> Joanne: inviting anyone to look at where provenance fits in tmo
  • <Bob> Joanne: where inf is captured, keep in mind in the use cases
  • mscottm i.e Richard Boyce described a knowledge base that could benefit from LODD and contribute a test case to provenance (sub)task force