Cory presents on open access publication and peer review… holy cow! Then he gets to rest and listen to Ruba and Matt.

Quick note: during Q&A, Ruba observes that maybe we miss important phenomena by focusing on reviewing all the A-lit and not reading the trade journals.

Ruba develops ontology and model management further (refer to earlier presentation).

  • Based on existing semantic Web tech: use what you’ve got
  • Standardization is a big deal! allows widespread adoption
  • Remember Krishnan and Chari (2000) give the really good background on model management
  • distributed model management needed (models just like info; gotta be able to share)
  • model management and Service-Oriented Architectures synergy
  • models are like services: they have inputs, processes, and outputs
  • ontologies are applied to services… maybe do same to models?
  • RDF: Resource Description Framework: mechanism for describing Web resources; effort at standardization of metadata
  • then came RDF-S, OWL (more complex), SWRL/SQWRL (these are pronounced swirl and squirrel)…W3C can explain further
  • a lot of this is about helping machines reason through rules, figure out things like “The brother of Bob’s son is another one of Bob’s sons.”
  • We’re trying to incorporate semantic info (that’s hard!) along with syntactic info (that’s easy) and apply it to decision-making models
  • Three levels, from abstract to concrete:
    1. modeling paradigm: capture constructs
    2. model schema: model indep. of data
    3. model instance: particular problem with data
  • Again, the service-model analogy is a big deal. Models can be thought of as services, services can be thought of as models….
  • [Warning: the acronyms are growing to, in one instance, six letters long. Could the length of acronyms could be an indicator of ripeness for a paradigm shift?… check that: eight letters. And a hyphen.]
  • I ask whether there is a field where we might get good empirical validation of the model; Deokar mentions semantic wikis in academic setting, useful also as portal for industry to register their own models in collaboration with other organizations, maybe security….

Matt Wills returns with electronic health records (EHR) and performance assessment (see Feb presentation). Tonight: some further investigation and proposal

  • Objectives:
    1. estab. strong theoretical foundation for model of clinical decision task performance with EHR
    2. redefine EHR in context of ERP systems
    3. explore/further define clinical decision task and EHR characteristics
    4. propose research protocol for data analysis and collection
  • Task-technology fit (TTF) theory (Goodhue and Thompson 1995)
    • nutshell: IT will have pos. impact when tech cap. matches task demands
  • EHR as ERP?
    • third-gen (3G) similar in many ways: basically, EHR can be the ERP for health care industry, can cover much the same business process aspects as ERP; applying the ERP model makes sense
    • TTF proposed for eval of entire IS infrastructure, not isolated tech
    • unlike EHR, ERP extensively studied, can guide this research
  • Four major themese in clinical decision theory:
    1. Pattern recognition (lowest level: requires no specific depth of knowledge; experience-based; basis of majority of decisions)
    2. Algorithms/heuristics (rules of thumb still grouping like pattern recognition, but requires more knowledge)
    3. Event-driven (no existing rules of thumb or patterns help; focus shifts to immediate therapeutic options, not driven by diagnosis or deep info)
    4. Hypothetical deduction (highest level of deduction)
  • utilization may drop from Matt’s approach to the TTF model, since EHR ut. will likely be mandated: nothing to study there! Performance is the only outcome variable of interest