This wiki has undergone a migration to Confluence found Here

October 26, 2010 CBCC Conference Call

From HL7Wiki
Jump to navigation Jump to search

Community-Based Collaborative Care Working Group Meeting

Back to CBCC Main Page

Attendees

Agenda

  1. (05 min) Roll call, call for additional agenda items & accept agenda
  2. (55 min) New project - Semantic Health Information Performance Standards

Minutes

1. Action Items - none

2. Resolutions - none

3. Updates/Discussion

Semantic Health Information Performance Standards

Today’s meeting focused on gaining feedback to a draft document sent to the list by Richard Thoreson and for getting a conversation going about a potential new project for CBCC. (Possibly joint with Security).

  • Attendance was minimal due in part to the fact the previous Security hour was canceled (but we achieved quorum).

  • Richard began the discussion saying that this draft is an attempt to develop an idea to help people to do what they should do, rather than trying to prevent people from doing what they shouldn’t. To date, our focus has more or less been to maintain controls to prevent bad behavior.
    • He suggests that we might be able to make use of the Security & Privacy ontology in the service of the most critical community and government- based policies (US realm) in the broader context:
      • how to articulate policy – not making policy – but create tools for making policy and tools for enforcing policy – access related to access control to health information based on policy.
  • Richard emphasized the considerable success that the creation of Domain Analysis Models (DAM) within HL7 has shown, in particular, the effort to create the Composite Security and Privacy DAM proved to be a useful way of pooling expertise and sharing information across areas of interest.
    • The proposal presented today is to create a DAM focused on reasons why people (patients and providers) might want (or not want) to share (sensitive) health information with entities not directly responsible for the direct provision of health care (research, governmental agencies, etc.) as well as to capture the information related to reasons why these agencies might want these data (including anonomyzed data).
  1. Public health and safety
  2. Prevention of costly, chronic health problems
  3. Early intervention when acute problems are most treatable
  4. Quality of care assessment and process improvement
  5. Setting spending priorities when funds are limited.
  • Richard has been in conversation with John Ritter about the PHR Functional Model and he is interested in talking about not only functionality, but the semantic meaning of data, so that it can be used for various purposes, in particular cost containment, quality and the issues that are driving change in the US.
  • The draft paper introduces the Proposal, an Approach and Target Information Performance Standards, but the intent is not to boil the ocean, and we’re not talking about what the policies should be, we are talking about the issues that policies have to address.
    • This project proposes to create a framework for tackling this technically: to define the information necessary to make an informed policy. The way ontologies have been used in the past has been more narrowly focused.
    • Can we identify the technical things that are not dependent on policy that but that can help governments to obtain the information they need to make policies that can impact finance, cost, etc? You can’t fix what you can measure, and you need data to create performance measures.
    • The quality improvement methodology being used in quite a few places is a context in which to think about how the ontologies might be built because you are looking for in some cases small but significant differences in behavior and performance that becomes your clue as to what changes could be helpful to gain further improvements in behavior and performance. There are natural variations that you might be able to surface with these differences in mind.
  • This is the larger picture of what Richard is trying to get at through the use of the ontology — how healthcare policy choices impact these relevant goals, without making assumption of what that the policy should be. You can’t fix what you can’t measure. Any reaction?
  • Jon Farmer: I’d like to rephrase some of those things back to see if I’ve understood.
    • If some of the stakeholders (agency, entities) beyond the day-to-day operational care provision for patients can make the case for the improvements they could make in the cost and delivery of care if they had this data in a way that is visible to path patient community, it’s conceivable that the patient might choose to share.
    • Step #1 is incentive engineering.
    • Step #2, with respect to boiling the ocean, you would avoid formulating policy, but instead would put into the DAM constructs to enable the evaluation of policy. So we’re trying to find parameters that any policy might need without getting buried into policy itself.
    • Step #3 – in order to translate this into a full feedback loop that improves quality and cost of care, someone who is doing the evaluation would have to choose what constitutes the key indicators, some performance or outcome number. How close is this paraphrase?
  • Richard: Good paraphrase. I might decide/encourage to share information if I know it’s something good, doing something good.
  • Richard: Through the use of this ontology, in a fairly concise and systematic way, we can explore the (technical) relationships between terms used in policy making but give them a more precise meaning. You could even ground the policy debate this way.
  • Serafina: Are we looking only at the Security & Privacy implications of the ontology with respect to this project proposal?
  • Richard: If you automate a process for sharing personal health information based on policies, you have to deal with access control, and it’s too complex if this process isn’t automated. We need to understand the technical information resources that you need to make well informed policy decisions.
  • Jon: What you’re talking about is defining the parameters for an access decision or sharing operation.
  • Suzanne: Can we define these parameters? For example, for every policy you need, X, Y and Z in order to evaluate it automatically. Policies need to be written in a way so that an evaluation can occur - policy supporting constructs (i.e., parameters) used for evaluation and formulation and to make these parameters available for implementation. This is similar to something I am working on with OASIS – related to the reference model. We’ve been recently talking about defining the information required to make policy available in a format that implementers can use. To accomplish this, we need to define a vocabulary or ontology that is common to all users.
  • Jon: There are two levels, and abstract and a concrete – a model and an implementation, and I imagine there are already policy definition languages out there.
  • Suzanne: Any examples of these?
  • Serafina: I thought that XACML and SAML were examples of security assertion policy languages.
  • Richard: Jon is referring to natural language policy languages, aren’t you?
  • Jon: I think XACML is a syntactical expression of some higher level constructs. Somewhere there has to be abstract thought structures, a conceptual language before you turn it into XACML. It’s not obvious to me that policy maker language is formally translatable, let alone automatically translatable into XACML for instance.
  • Suzanne: What I was talking about in reference to the OASIS project is the need for a policy structure. Not creating the policy, but the policy itself needs a structure, so that whatever is going to evaluate the policy, a format (syntax) is necessary.
  • Jon: A lot of policy would be expressible in a construct like this: such and such an operation is permitted if X, Y & Z. And X, Y & Z can be Booleans that can be evaluated. So, by what constructs can those individual predicates (booleans) be formulated? It is dependent on context, content and purpose. I don’t know if any of the existing description languages actually distinguish between those three things and what those structures would look like.
  • Serafina: Perhaps I have misunderstood what we’ve been talking about all along with respect to XACML and SAML – that the parameters in these languages have been identified and are relevant/important for evaluating security (and privacy policy), and that what is needed is an ontology that enables all parties involved in an exchange to evaluate/understand the meaning of the terms (used in these parameters) to make accurate access control decisions.
  • Richard: We’re trying to tease out the complexity of the many dimensions of these terms (e.g., Purpose of Use).
  • Jon: If I hear you correctly Richard, it’s one thing to put up a list of ten options for Purpose of Use for selection; it’s another thing to take those selections – what do you compare them to in the operational environment?
  • Richard: But first you have to explain what those ten options mean before they can be selected. We need a framework to establish ground rules in what would be a potentially contentious environment. If we can agree on definitions, we can carry on a reasonable conversation.
  • Suzanne: So a question to you Richard, are you proposing a new project? So do we want to create a project scope statement from this draft?
  • Richard: Yes, but someone has to come up with resources...
  • Serafina: But doesn’t this relate somewhat to the Privacy Policy Template Catalog project that we kicked off a while ago and which is on hold due to lack of resources/funding?
  • Richard: This is true.
  • Suzanne: Let’s think about extending the work that Steve Connolly started a while back and which we are now incorporating to some degree in the Security & Privacy Ontology where we are mapping existing standards to the S&P Domain Analysis Model, where we are in effect creating the “silos” to harmonize the standards across the board for our DAM and ontology. The CBCC WG will develop the context and purpose, but use the data that is coming out of the mapping being done in the mapping of standards to the Security & Privacy DAM. We can use the content that is being teased out of the standards mapping to use for the content for our needs. We would run this project in parallel to the Security project (mapping).
  • Richard: My concern is that this will be seen by the Security folks as boiling the ocean. We’re just trying to use the concept of policy in a much broader context that it has been defined in the Security space. But I think that the larger policy context (government, community policy) is going to have to be factored into access control (security policy) algorithms going forward.
  • Jon: I agree that what Privacy and Security has been doing in computing history has been more coarse-grained and we’re venturing into more fine-grained area of content discrimination and this may be scary to many. But if we’re going to give consumers fine-grained consent privileges, you’re not going to be able to automatically enforce it unless we go here.
  • Richard: This is applying content-related policies, the dimensions of policy – scope of information, accuracy, precision, timeliness – it is going to be really important in assessing the sensitivity of information in EHRs, PHRs, etc.
  • Jon: Exactly, the existing object catalog in the RBAC spec, those are not really down to content, they have more to do with packaging – like discharge summary. That’s not talking about what may or may not be sensitive. We would delineate some bodies of content on which you could parameterize some policies – it’s not the whole universe. It keeps it manageable.
  • Richard: The key dimensions of quality: strength of privacy and security controls, transparency and ease of use of access controls, clear and effective anonymization and pseudonymization protocols – these things that are going to impact cost and effectiveness are going to become quite controversial. It is going to be important to work through the ontology in a way that people on both sides of any discussion can agree to. I’m hoping this isn’t jumping the gun too much. But at least in the US realm, how these terms and things relate to each other is vital.
  • Serafina: I think Richard’s draft and the discussion it can generate is very relevant to the Security & Privacy ontology project because it provides a basis (real-world considerations) for those of us who do not think as abstractly as others, for participating in what is a rather abstract endeavor (defining the ontology).
  • Jon: Policy is totally abstract in reality – a rule for what may or may not be allowed. There is nothing concrete about it. So you have to agree on these abstract constructs.
  • Richard: One of the challenges is resourcing, but also, identifying the low-hanging fruit. What part of the domain, are most likely to be manageable in terms of the ontology and directly relevant.
  • Jon: Do you have any small pieces in the Substance Abuse and Mental Health domain that you would find useful?
  • Richard: If we could figure out a way to break down what is a big job into smaller, manageable chunks. I will think about this. I first need to sit down and work with Apelon to show the relevance to Behavioral Health, but I would like to have Mike and others participate in these discussions since there is relevance to the VA and other federal agencies.
  • Suzanne: The VA has a lot of interest in Privacy of course.
  • Richard: The objectives need to be well articulated to gain support for this kind of work. But to achieve automation of fine-grained consents, we need to continue to try to work through these issues.

We reached the top of the hour. We will continue this discussion in next week’s CBCC hour, hopefully with a larger group of participants.


The meeting was adjourned at 3:04 PM EDT. No significant motions or decisions were made.

Back to CBCC Main Page