CCOW Meeting Minutes – 5/23/00 Minutes recorded by Jon Farmer and Mark Morwood Rob Seliger Brian Bishop Mark Morwood Gunnar Klein Melvin Reynolds Jon Farmer Barry Royer Mohamed Oukani Paul Biron Cameron Moore Andrew Woyak Gary A. Bartlett Michael Russell Luis Edwards Kathy Kemper Hennie du Plessis Ed Larson Sentillion University of Alabama at Birmingham Health System Sentillion Karolinsua Institute, SW AMS Consulting Care Data Systems SMS IDX Keiser Permanente 3M MedicaLogic/Medscape Emory Healthcare Duke Univ Health System Gettysburg Hospital Group Health Co-op PQ Africa, South Africa E R Larsen Inc robs@sentillion.com bebishop@uabmc.edu markm@sentillion.com gunnar@klein.se melvin_r@amsc.demon.co.uk jfarmer@ic.net barry.royer@smed.com oukani@idx.com paul.v.biron@kp.org ckmoore@mmm.com andy.woyak@medicalogic.com gary_bartlett@emory.org michael.russell@duke.edu ledwards@gettysburghosp.org kemper.k@ghc.org hennie@qedi.co.za elarsen@erlinc.com Discuss CCOW web-specification negative Ballot issues Discussion on use of term W3C vs. “web” in CCOW web-spec.; decided to simply search and replace with “web”. Likewise, for Negative Ballot pg.28 line 1 suggestion to use “web” rather than W3C. Will give Unicode codepoint values to define characters. Agreed to add document references rather than (or in addition to) summaries Agreed to take no action on overly-conservative inclusion of hex characters P37, line 10-11: Characters set: replace each “Unicode character” with “Unicode codepoint” and strike the “(UTF-8)” P9 line 1: correct by removing statement that connections are broken and reestablished for each HTTP message. Unicode vs. ISO xxxxx vs. UCS: will say “Unicode”. URL vs URI: URLs are locators. URIs are generic identifiers (and are not necessarily dereferenceable to find their location), URNs are names. We should be using URLs not URIs (and we are). We agreed our spec is concerned with “Absolute Dereferenceable URLs”. Will use URLs to the exclusion of URIs. Section 6 first paragraph will be clarified regarding requirements of posts and gets. Side-effects issue: \ CCOW is using native HTTP Gets and POSTS as a remote procedure call mechanism, but is subject to concerns regarding caching side effects. We agreed to require use of “nocache” parameter on both gets and posts so as to eliminate vulnerability of receiving cached responses. URL encoding of literal CCOW as part of filepath – agreed NOT to required it. Method names – parsers should ignore parameters they do not recognize. Also remove the implication of a required order of parameters, since they are named Web Spec comment resolution continued: ParticipantCoupon needed for secure get: Agree – Arch spec now state as optional dependent upon mapping. Web requires. CP URL as data for CMR::Locate. Agree. Clarify the use of the URL returned by CMR::Locate, Resolution: return two args, the base URL plus optional args. These must be appended to the URL or sent as POST data. Dependent subjects: clarify wording. DAGs allowed. Directed Acyclic Graph. Vote in favor of accepting changes 9 yes, 2 abstain. (Paul Biron agreed changes met his requirements and will send an email changing his vote to affirm) CCOW 1.2 is passed! CCOW 1.3 goes to full ballot. Discussion: CCOW uptake. Lunch. Review CCOW Roadmap from last April and compare with current status. Ideas discussed: CCOW Longitudinal View, Summary View, … (John Farmer) CCOW + wireless XML SOAP Role-based access privileges Multi-device context sharing Multiple concurrent context sessions Sharing context across a co-located workgroup (e.g. herd of doctors on rounds) Trusted environment context ACLs for Context Subjects Session management (SMS), single sign off, activity monitoring Context Events Workflow Context (Macro/Micro) Person context – (edu person – U. of Michigan; c.f. RIM’s person object) Stacked Contexts Addition contexts HIPAA For tomorrow assess next steps based on some criteria: e.g. Simplifies app integration Providing end-user features Leveraging existing CMA Time to market Backwards compatability Leverage/complement existing standards Agenda for tomorrow: CORBA technology mapping draft SOAP technology mapping Roadmap discussion and refinement Action plan for next meeting Minutes for CCOW session Wednesday 5/24/00. Minutes recorded by Jon Farmer and Mark Morwood Attendees: Barry Royer Gary A. Burtlett William Eckels Simon Curtis Mark Morwood Jon Farmer Michael Russell Melvin Reynolds Cam Moore Alan Chan Andrew Woyak Mohamed Oukani SMS Emory Healthcare IDX Systems Corp IDX/Channelhealth Sentillion Care Data Systems Duke AMS Consulting 3M VA Medicalogic/Medscape IDX Barry.royer@smed.com Gary_Burtlett@emory.org William_eckels@idx.com Simon.curtis@channelhealth.com markm@sentillion.com jfarmer@ic.net Michael.Russell@duke.edu Melvin_r@amsc.demon.co.uk Cmoore@mmm.com Alan.chan@med.va.gov Andy_woyak@medicalogic.com oukani@idx.com CCOW CORBA Technology Mapping Mark Morwood presented draft spec created by T. Haug and A. Klingler CCOW SOAP Technology Mapping start mapping work when market looks ready and when somebody wants to work on it Will assess benefits – are there particular benefits, or, “just another mapping” SOAP implementation would likely might be be less complex than the HTTP mapping. Maybe will permit integrating the component’s security infrastructure by holding chain of trust Function Activation A high-demand application of this is for participant to ask for the ”patient search” function to be surfaced. o Not clear whether CM will delegate or just give the caller a pointer o Apps register the functions they implement (like patient search) o Any participant can activate the “patient-search” app through the context manager. This is powerful because the participant would not have to know which app can do the search. o Many departmental systems would rather not implement patient search themselves (radiology system doesn’t want or have the whole patient population in its local MPI) Also would be useful for an app to ask for activation of the user authentication. In both cases, the common concept is that a participant is asking for an instigator for a subject. We should still keep it generic and extensible for use with other services. Use Model for Multiple Device, Shared Contexts, Stacked Contexts Presently, the device and the context have a 1:1 relationship. How to support multiple contexts on a device? How to adequately decouple the devices from contexts? PDAs might be an important device to address soon – to get context from a workstation, or PDA patient gets set from bed; But the PDA should set patient as approach monitor at nursing station. Probably need to specify rules. While approach device – join its context and then instigate or be instigated, by subject. A location context subject? With dependency rules? Does the mobile device have its own context (at least user – so user need not reauthenticate as he walks to different device) How does a device put a location into context? And it persists (or reinstates) when the PC is turned off. What does a “location” look like? A Facility/Unit/Room/Bed might be an 80/20 starting point that specifically addresses the use case of caregivers approaching beds. How will the bed context change drive the patient change? How to coordinate multiple context managers? Sharing subjects? Physician (user) can be in two organizations and two contexts Multiple concurrent contexts for one user in multiple enterprises (e.g. doctor doing referrals) enterprise Context “stack” for interruption-contexts Allow apps to store any application state to be restored when appropriate Multiple concurrent contexts with shared and unshared subjects – is this a context item with multiple instances? Or a separate context? When joining, how does a participant specify which subjects of which contexts? Context spawning – inheriting a “starting context” with the same user. Usability issues – determine differentiation and establish defaults. – Maybe use color and fonts. Barry volunteered to do a white paper on mult-context-single-device Single Signoff - means set user context to null ACLs for Context Subjects – Private subjects – that only certain apps can see Since Confidentiality policies are not commonly expressed in terms of which apps can see what, we are not sure this is adequate popular demand, or that it is an appropriate HL7 problem to solve. Trusted Environment Context Those present didn’t feel qualified to discuss this one without having the idea-originating person present. Context Events Publish & Subscribe – not significant priority for more dicussion today Workflow Context Sync all apps to the “todo” lists? User’s workflow? Sounds like an application. Person Context As opposed to patient or user – we think we should address this when we do V3-RIM-based subjects or stakeholders Additional Contexts Example – be able to propagate current insurances for the selected patient. HIPAA Given user context and and patient context – useful to enforce access controls? Session is the user plus the consent policy CCOW provides security parameters (user identities and roles) to apps Security – Authentication, Auditing of Accesses as a context trail Identifiers – mapping agent can permit national identifiers to be moved into the context for the apps. Groupings Security Context Complication S HIPA e A c M H PRIORITY CCOW + wireless Longitudinal View XML SOAP Role base access x x privileges Multi-device context x sharing Multiple concurrent contexts Context sharing x across workgroup Trusted envt context x ACLs for Context x Subjects Session x Management, single signoff Context events Workflow ctxt Person context Stacked context Additional context HIPPA x x Dynamic context x definition (Ken Rubin) *Subject to stability of RIM items More Complex Context M Device Support Tech mapping M x L x XML / v3 H* Funct Activ ation M x x x x x x x x x x x x x x Next Steps Due to the lack of a quorum after lunch, it was decided to establish follow up action items for the next meeting using email to the CCOW list. Before he had to leave to catch a plane, Barry indicated interest in producing a white paper for the next meeting on the requirements and issues associated with supporting multiple contexts per desktop and multiple desktops per context.