Wednesday, June 29, 2016

Apple makes a difference with targeted use of Differential Privacy

Amazing news this week from Apple on their use of Differential Privacy. They didn't invent Differential Privacy, they are not the first to use De-Identification, but they seem to have thought about this. We can't yet tell how well they thought about this, as the details are not clear. That is, they are still not being Transparent.

I am not an Apple fanboy, I think they get far too much credit and buzz. I don't think this is their fault, they are masters of marketing, and they never claim to have created something no-one else has. What they do very well is take technology that is just slightly behind bleeding edge, letting someone else get most of the cuts and blood, and the critical thing, they use it in a way that provides really good value. It is this that I am very much a fanboy for apple, they know how to pivot the work of others into a bigger value for their customers.

Differential Privacy is another case of this value-adding exceptional execution.   Specifically...

Wednesday, June 22, 2016

On-Behalf-Of - FHIR Signature datatype update

The Security workgroup is looking at the FHIR datatype for Signature, specifically the use-cases where the one signing is not the one that the agreement or contract is about. For example when a parent signs for an infant child, or a guardian signs for an individual, or where an individual signs for an organization.

We look to the Uniform Commercial Code (UCC) as one use-case for the use of On-Behalf-Of, as it is a long-standing and proven case where this is needed and has been used. ....


Tuesday, June 14, 2016

MHD - Why use of FHIR Contained?

I was asked why MHD has the requirements to contain a provider and patient in document reference. Why is that necessary?

The containment of Provider and Patient are ONLY for very specific reasons. Some are driven by XDS constraints; but even those constraints are driven by reasonable use-cases.

Friday, June 3, 2016

IHE #FHIR profiles - MHD, PDQm, and PIXm

The IHE IT Infrastructure committee has a few profiles on #FHIR. They just published new revisions  (see below) that have aligned on DSTU2, formerly on pre-DSTU2 versions. They have also just published for Public Comment a new set of supplements. All of these are aligned with DSTU2. When STU3 finishes ballot and publishes reconciled version, then IHE will adjust all of these to STU3.

Sunday, May 29, 2016

Simplified #FHIR Privacy Consent Directive resource

The most simple Privacy Consent Directive would really just be a YES/NO flag, so I don't actually mean that simple. What I mean is more simple than the last update I made, yet still functional. With the previous version I learned a few more things to be simplified.  The most shocking that I forgot is that the most prevalent exception is to exclude data published duing a date/time range.

So now I have updated the Privacy Consent Directive resource that has a base that identifies the patient, authority, domain, location, recipient, grantor, data, and actions. These are the elements needed for an all-or-nothing kind of consent.


Simple FHIR Consent

Much discussion around consent (Privacy) in HL7 FHIR. It seems that much of the seamingly endless circular arguments come down to getting everyone on the same page.

1. HL7 can only produce Interoperability standards. It can't define Policy, Architecture, or Implementations.
2. We thus can't just pick one architecture (e.g. OAuth+UMA) and stop. We must build to fit a reasonable set of architectures.
3. We thus can't jut pick one policy (e.g. patient home) and stop. We must build to fit reasonable a set of policies.
4. We thus can't just pick one implementation (e.g. single server) and stop. We must build to fit a reasonable set of implementations.
5. FHIR should focus on only what systems do today, not what we want them to do.
6. FHIR should put into common extensions those things that we want them to do (e.g. Identified disclosure notification endpoint)
7. Simple is always better

This means that sometimes a specific Policy, Architecture, and Implementation might not need our solution at all. This is likely the case if someone picks the HEART (OAuth+UMA) model, were the current FHIR Consent adds no value since all the consent policy management is handled inside the UMA authority.

Adrian 'working defintion of consent' is much like what we have used. I reproduce it here with attribution as I like his structure:

I (Adrian) propose a working definition of consent as:
  • A contract between a person and an institution that describes a policy of the person and a duty of the institution.
  • A policy is not an attribute. An attribute can be measured and asserted by a lab or a nurse. A policy can only be asserted by the patient or their custodian - hence, the need for a contract.
  • A contract combines a series of terms into one accountable transaction. 
  • The terms of a consent transaction MUST include:
    1. An institution that carries the duty of interpretation.
    2. A subject - the patient
    3. A custodian who is allowed to assert a policy (the patient or legal guardian of the patient)
    4. A resource (potentially any non-aggregated FHIR resource, including de-identifed person-level resources)
    5. A policy to be interpreted by the institution related to use of the resource (this will become a scope for authorization to access the resource at some future time).
    6. A notification endpoint chosen entirely by the custodian for transparency in how the institution is interpreting the policy to issue each authorization.
    7. A signature of the custodian
    8. A signature of the institution that will issue the authorizations
    9. A jusrisdiction for the contract,
This definition of consent is based on my amateur interpretation of international Agency Law. If we can agree on this, then the role of FHIR in capturing consent would be immensely clearer.

I think this maps well to what I have already created in the FHIR Consent resource:
I think you will find it matches quite well. With a few exceptions that we have not yet thought of (thanks for bringing them up), or things we might put into extensions due to the 80% rule.


That said, we can always improve. I welcome these constructive comments. I welcome suggestions for better name of the Resource (Consent) and better names and definitions for the element names.

I have already realized through discussions with Adrian, Oliver, Rob, Kathleen, and Lloyd that the model I have could be further simplified, extended, and clarified. 

First generation FHIR Consent Scope

I would like to assert that we are focused on enabling a reasonable set of policy architectures and policy sets. What we create must degenerates nicely to support environments that want a simple indicator of opt-in, opt-out, none. But we want to go a bit further, we want also to support opt-in with some reasonable exceptions, and opt-out with some reasonable exceptions. Where these exceptions are only those currently implemented: List of objects, List of authors, List of recipients, List of Organizations, List of purposeOfUse, and Date Range. --- This scope is however not properly said so far, but I would be glad to indicate it or a sub-set as the goal.
Improvement Opportunity:
1. Simplify Actor to the specific kinds of actors in a consent. using specific elements, not a list with type.
1.a) Be specific about recipients, and allow device to be a recipient. Where recipient is a named identity of something that is given authority under the consent.
2. Simplify the attachment to just one, no need to differentiate between legal and friendly today. Maybe in the future
3. Simplify to remove the rule element. Possibly move it to extension
4. Add exceptions for date-range
5. Make the exception more clear

Saturday, May 28, 2016

Public Comment period for IHE Advanced Patient Privacy Consents profile

IHE IT Infrastructure Technical Framework Supplements Published for Public Comment

The IHE IT Infrastructure Technical Committee has published the following Technical Framework Supplements for public comment in the period from May 27 to June 26, 2016:
  • Add RESTful Query to ATNA
  • Advanced Patient Privacy Consent (APPC)
  • Re-documentation Register On-Demand Document Transaction
  • Mobile Alert Communication Management (mACM)
The documents are available for download at http://ihe.net/Public_Comment/. Comments received by June 26, 2016 will be considered by the IHE IT Infrastructure Technical Committee in developing the trial implementation versions of the supplements. Comments can be submitted at ITI Public Comments.