Concall Scientific Discourse Task Force November 21 2011
Tentative use cases as per 21 Nov 2011
- Status report on three use cases we are working on:
- A linked open data store of scientific information that updates and elaborates on medication safety statements present in drug product labels Team: Richard Boyce, Anita de Waard, Maria Liakata, Jodi Schneider, Mike Taylor
- Defining minimal information requirements for biomedical investigations defined and mapped to RDF - Team: Tim Clark, Susanne Santone, David Shotton
- Mining Treatment Response Patterns and Outcomes - Team: Joanne Luciano, Bob Powers, Anita de Waard
- New Format/schedule for these calls
Notes #hcls2 HCLS Scientific Discourse call, Mon Nov 21 2011 10:35:13 GMT-0500 (EST):
Attendees: Richard Boyce, Joanne Luciano, Jodi Schneider, Tim Clark, Eric Prud’hommeaux, Anita de Waard, Mike Taylor, Matt Gamble
- Use cases.
1.1. Use case 1: "Dynamic enhancement of drug product labels through semantic web technologies" - Leads: Richard Boyce, Maria Liakata, Jodi Schneider, Mike Taylor, Anita de Waard https://docs.google.com/document/d/1QpW-axtGL7Tuhd_Zcaf30a4-s4S_lveIJtpKL5QBLfQ/edit?hl=en_US
- Use case is more complete, individuals are revising specific sections in Google Doc. Meeting in 1 hour to further review.
- Title "A linked open data store of scientific information that updates and elaborates on medication safety statements present in drug product labels", may change to Michel's suggestion, "Dynamic enhancement of drug product labels through semantic web technologies"
- motivated by helping reduce bad drug interactions, use technology to identify claims in package inserts & sci lit to update these
- develop a Linked Data store to update this info
- narrowing to age-related changes in clearance, pharmacokinetic drug-drug interactions, and metabolic clearance pathways
- See here for a presentation-of-the-PI-mashup-use-case-11032011.pdf
- Narrowing to antipsychotics, antidepressants, and sedative hypnotic drugs /sedative hypnotics -all under psychotropics
- Maria Liakata and NacTEM (Manchester, National Center for Textmining) - very interested in finding key side effects/adverse affects in the literature -they would like a very bare-bones ontology of claims statements, possibly using AO -Annotation Ontology - http://annotation-ontology.googlecode.com
- ORB as a way to narrow down WHERE we're looking in a document- http://www.w3.org/TR/2011/NOTE-hcls-orb-20111020/(ORB is basically IMRAD as an ontology)
- Any questions?
- Joanne Luciano: How do we know when it's done? What outcomes are expected?
- Rich: Generate a view of the package insert. Test it with users (who maintain Hypocrates/drug databases/etc). Want them to see what's out of date in the package insert, context in the sci lit that should be assessed.
- Matt_Gamble: Is there any plan to deal with quality & trustworthiness of the Linked Data?
- Joanne Luciano: How and who is looking at provenance? It would be important in this use case.
- Rich: Could use for Structured Product Labelling (HL7 standard).
- might also want to consider IAO - information artifact ontology : code.google.com/p/information-artifact-ontology
- Anita: Patient orgs can deposit/use the data.
- JoanneLuciano: controlled environments vs. real-world
- JoanneLuciano: want it to be more user-friendly than just a SPARQL endpoint, want to empower people
- Boyce: we should brainstorm and think how can this be relevant for the patient? But want to direct this first at intermediaries, then at patients.
- We’ll invite Joanne Luciano to a next meeting.
- Matt_Gamble: Interesting article w.r.t open drug data and quality from Antony Williams: http://www.drugdiscoverytoday.com/download/617/
1.2. Use case 2: "Defining core metadata for describing biomedical investigations" Leads: Tim Clark, Susanna-Assunta Sansone, David Shotton, Philippe Rocca-Serra http://www.w3.org/wiki/HCLSIG/SWANSIOC/Actions/RhetoricalStructure/meetings/20111031.
- Tim describes current work:
- build OWL framework, using work we'd already done.
- David's concern: minimal info models, LOTS of them. Minimal info model for small subdomains, e.g. parasitology. David Shotton's observation: none of these models were on exactly the same page, no core model connecting them
- David's goal: core, minimal info model for describing biological investigations.
- Tim also had meetings with Manchester, Oxford, Susanna, David, open access publishers -- interest in *using* such a core model. Question: do we attempt to define a core model? Or work stepwise?
- New stuff (used in stem cell;) "factor"--in design of a study. biomaterials & tech platform. Factors are independent by subdomain - need for both factors model and trust model.
- Perhaps 2 use cases, not just one
- Separating concerns -- info quality still needs to be done
- ISA RDF model will still be done. Would like to align with other aspects of this work, too
- visiting David Karger till June next year, from Carol Goble's lab at Mancester.
- Coming at minimum info models from info trust & quality on scientific data aligning various models and ontologies. Minimum%20Information%20Model%20Vocabulary%20Specification.pdf
- From paper checklist --> structured representation- 177_paper.pdf <- an example paper - kind of the grand idea of my work
- instrument information is also really important –
- TIM: you are very right
- Student is doing a thesis on explanation –to support explanation in the bionmedical domain
- Anita: interesting, please send details!
1.3. Use Case 3: "Matching patients to treatment outcomes" - Leads: Joanne Luciano, Bob Powers, Anita de Waard
- JoanneLuciano: better matching treatments and patients, on a biological/psychological basis
- Have treatment decisions be more of a science than an art!
- Black box model (responders/non-responders), white box (underlying dynamics)- patterns of recovery (change in order of indiv. symptoms) different for different treatments
- 2D grid: how popular vs. how effective. 98 different treatments.
- Also include patient information mined from the web
- We need to meet on this and decide what to do. Invite Richard Boyce to the meetings.
- Richard Boyce is interested in connecting to people at Pittsburgh:
- Judith Callan is the researcher here at Pitt working on CBT and smart phone apps: http://www.nursing.pitt.edu/faculty/facultydetail.jsp?primkey=callanja
- for example, http://journals.lww.com/nursingresearchonline/Abstract/2011/09000/An_Intervention_Fidelity_Framework_for.9.aspx
2. New format for HCLS call: let’s have monthly call with quick status update (like this one), may also invite speakers and use the other 3 slots for the other 3 use cases, in succession?
- monthly call with speakers is a good idea; however perhaps we might not be all on the same page; chairs would need to play a larger role in being the connective tissue
- Anita: ok I’ll send out a tentative schedule + request for speakers
- Jodi :
- Whatever happened to this use case? “Structuring computer science publications during authoring “-Leads: Tim Clark, Tudor Groza
- Anita: I don’t think we had the time; someone should talk to Tudor, time difference is a problem – Tim?
- Someone has to do something with these minutes
- Anita, two weeks later: hereby…