[Openid-specs-heart] HEART stepping stones

Mon Apr 20 16:37:32 UTC 2015

It may also be helpful to consider the transparency perspective. Regardless
of how many institutions lay claim to a patient-specific resource, there's
only one patient-subject. I'm not a security expert, but it seems to me
that regardless of how the access decision is made and how it turns out,
good security practice suggests that the subject be notified that something
happened and that she has the opportunity to seek details and / or manage
similar notifications.

Adrian

On Mon, Apr 20, 2015 at 11:44 AM, Justin Richer <jricher at mit.edu> wrote:

> For my two cents:
>
> I’ve long said that while the resource server itself is always the
> ultimate authority on data flow (since it can blissfully ignore any other
> external inputs), there is a calculus of access rights that needs to occur
> before an authorization decision can be reached. This calculus is
> multivariable in nature, with both several inputs and outputs.
>
> The UMA/OAuth access token is one such input variable. It contains aspects
> such as the scope or permission set, the authorizing party (user), the
> authorized party (client), and others. This is a powerful construct that
> can cross security domains. With the right tools (like the HEART profile
> will define) you can determine who made the token and who sent it and
> whether or not you care.
>
> But it’s not the only input. The RS could have many other tools at its
> disposal, and in the healthcare space it likely will. You’ve got heuristic
> analysis, policy servers, firewalls and gateways, etc, all working along
> side the access token itself. So if a user get an access token from their
> personal authorization server, the RS is still going to check its local
> policies. Are they allowed to access this record at all? Do they only have
> access on a limited timescale? Are they asking from a suspicious IP
> address? When was the last time we saw this user in person, anyway?
>
> And a yes/no decision isn’t the only output. Did the token tell me a set
> of scopes that would cause a privacy leak when used in combination? Is
> there another policy that would limit the information being sent in this
> response, no matter who’s asking?
>
> All of this needs to be considered and pulled together for the request.
> Much of this needs to be expressed in a way that’s centralizable and
> cacheable, for performance and administration sanity sake.
>
> At the end of the day, the user-managed authorization server is a very
> important part of this calculus and one of the ones that HEART is focusing
> on. But we’re not leaving out the organization’s authorization server
> either. We want to *enable* these different components in such a way that
> when they exist they can all work together.
>
> You see, all we’re really trying to do here is change the world for the
> better.
>
>  — Justin
>
>
> On Apr 20, 2015, at 10:40 AM, Debbie Bucci <debbucci at gmail.com> wrote:
>
> Hesitant to speak up but since John asked ...
>
> With regard to UMA Authorization Servers, are you suggesting that we
> consider a mix of personally-controlled and institutionally-controlled
> Authorization Servers or just one or the other?
>
> *Mixed.  I could see places where an Authorization service would/could be
> logically stood up outside an institutions borders (in case of Health IT -
> ACO, HIE etc).   Additionally if these entities focus on patient/consumer
> value add service, those authorization services could/should allow the
> patient to add additional end points ...perhaps even federate with other
> known/trusted authorization services.   Including Adrian's 5.00 a month
> service - providing the binding is strong enough to be trusted.*
>
> With regard to interface scopes, are there particular scopes that should
> be considered before others?
>
> *Don't understand this question.  I think its use case driven*
>
> With regard to identity management and identity federation, would we
> consider patient ID before or after provider ID?
>
> *In order to access the API the identity negotiation would need to be
> completed upfront. In the in PoF demonstration - we repeated said it was
> out of scope but if you looked closely ... Alice did use a federated
> credential.  John did bring up identity proofing/LOA/trust in one of the
> early calls.  Even though we do not deal with it directly it does need to
> be represented/addressed and is a necessary part of the
> authorization/access "calculus".     I know there are a number of folks on
> this list already tackling this problem space and are looking for way to
> integrate into these profiles/workflow.  We should let them help us.   *
>
> With regard to patient matching and discovery, would we try to keep these
> in or out of scope for the early parts of the roadmap?
>
> *If we presume the patient is mediating in the center and has a a explicit
> binding to their resources  - there are  no  matching issues. *
>
> *Client dynamic registration and AS discovery would be in scope from my
> POV.*
>
> *There has been a very promising discussion on the UMA list about a
> webfinger-ish personal discovery service.  Not real yet though- a gap that
> I hope get closed in the near future.*
>
> Is there a class of providers or data holders (hospitals, payers, labs,
> public facilities, etc...) that we could prioritize?
>
> *Do we need to prioritize?  Who's willing to share? Please  let us know!*
>
> *Separate concerns - *
>
> *If we believe the JOSE/JWT is essential for secure data exchange - we
> should stand behind it not compromise.*
> *If we unearth some real policy concerns (US and International) or gaps in
> the standards  - how do we place in parking lot/acknowledge for others to
> tackle.  Ae there folk on this list willing to take on some of those
> challenges?*
>
> *Deb*
>
> *P.S.  Disclaimer - Deb's personal views  mindfully sent using Deb's
> personal email.*
>
>
>
>
> On Sun, Apr 19, 2015 at 9:47 PM, Adrian Gropper <agropper at healthurl.com>
> wrote:
>
>> Then this is an excellent discussion. It suggests that there's a roadmap
>> and some metric for achievability.
>>
>> For example:
>>
>> With regard to UMA Authorization Servers, are you suggesting that we
>> consider a mix of personally-controlled and institutionally-controlled
>> Authorization Servers or just one or the other?
>>
>> With regard to interface scopes, are there particular scopes that should
>> be considered before others?
>>
>> With regard to identity management and identity federation, would we
>> consider patient ID before or after provider ID?
>>
>> With regard to patient matching and discovery, would we try to keep these
>> in or out of scope for the early parts of the roadmap?
>>
>> Is there a class of providers or data holders (hospitals, payers, labs,
>> public facilities, etc...) that we could prioritize?
>>
>> Adrian
>>
>>
>>
>> On Sun, Apr 19, 2015 at 9:33 PM, Moehrke, John (GE Healthcare) <
>> John.Moehrke at med.ge.com> wrote:
>>
>>> I am not trying to limit the destination. I am trying to define the next
>>> achievable step.
>>>
>>>
>>>
>>> John
>>>
>>>
>>>
>>> *From:* agropper at gmail.com [mailto:agropper at gmail.com] *On Behalf Of *Adrian
>>> Gropper
>>> *Sent:* Sunday, April 19, 2015 5:13 PM
>>>
>>> *To:* Moehrke, John (GE Healthcare)
>>> *Cc:* Eve Maler; openid-specs-heart at lists.openid.net
>>> *Subject:* Re: [Openid-specs-heart] HEART stepping stones
>>>
>>>
>>>
>>> Hello John,
>>>
>>>
>>>
>>> There's no need for you to take my perspective personally.
>>>
>>>
>>>
>>> "Data created fully by the patient" seems to be urging us to down-scope
>>> HEART to the non-HIPAA domain.
>>>
>>>
>>>
>>> Adrian
>>>
>>>
>>>
>>> On Sun, Apr 19, 2015 at 5:21 PM, Moehrke, John (GE Healthcare) <
>>> John.Moehrke at med.ge.com> wrote:
>>>
>>> Hi Adrian,
>>>
>>>
>>>
>>> Interesting misrepresentation of what I said. I am disappointed that you
>>> feel it necessary to misrepresent what I said. I am also disappointed that
>>> you feel it necessary to bring in other negative topics that I said nothing
>>> about. I am trying to find ground that we can progress forward on; while
>>> you seem to be just wanting to make personal assaults.
>>>
>>>
>>>
>>> Looking for the constructive message in your comment, I think you are
>>> suggesting that we scope our efforts to the flow of information from the
>>> patient possession to points-elsewhere. I am fine with that kind of a
>>> scope. It also avoids the issues I was bringing up.  I very much agree that
>>> data created fully by the patient is, and should be, totally controlled by
>>> the patient.  This scope also avoids the concerns that encumber healthcare
>>> provider environments:  Medical Ethics concerns, Safety concerns, and
>>> concerns of wrongful disclosure.
>>>
>>>
>>>
>>> John
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> *From:* agropper at gmail.com [mailto:agropper at gmail.com] *On Behalf Of *Adrian
>>> Gropper
>>> *Sent:* Sunday, April 19, 2015 12:42 PM
>>> *To:* Moehrke, John (GE Healthcare)
>>> *Cc:* Eve Maler; openid-specs-heart at lists.openid.net
>>> *Subject:* Re: [Openid-specs-heart] HEART stepping stones
>>>
>>>
>>>
>>> John, I find your perspective both paternalistic and unscalable.
>>>
>>>
>>>
>>> US healthcare is awash in lack of transparency and the result is
>>> $1Trillion of unwarranted care. It's paternalistic and incredibly
>>> self-serving to presume that just because the institution has been given a
>>> right to use patient data without any accountability as long as the data is
>>> for Treatment, Payment, or Operations or De-Identified, or "Break the
>>> Glass", or prescription drug monitoring, or just plain lack of segmentation
>>> for access, that it's good policy. The current regulations are the result
>>> of heavy and effective lobbying by a very well organized industry trying to
>>> protect its secrets by avoiding the HIPAA accounting for disclosures and
>>> and patient right of access because they're "too hard". Think of HEART as
>>> trying to fix the "too hard" problem.
>>>
>>>
>>>
>>> Your perspective is also unscalable as more and more health-related data
>>> originates in wearables as well home and environmental monitors, and then
>>> ends-up in trans-national analytics completely outside of the HIPAA regs.
>>> It's also unscalable as patient data such as genomes can no longer be
>>> collected under informed consent because nobody has any idea of how your
>>> genomic information will be interpreted three years from now and how that
>>> interpretation might affect you or your children. It's also unscalable as
>>> the ability to promise de-identification for research becomes less and less
>>> realistic.
>>>
>>>
>>>
>>> The simple fact is that surveillance, data processing, and data storage
>>> is now effectively free compared to the economic value of the patient data.
>>> Rent-seeking-behavior by politically astute institutions has been effective
>>> for the past few years but the natives are getting restless. If you want to
>>> read more:
>>> http://thehealthcareblog.com/blog/2015/04/16/last-chance-for-meaningful-use/ and
>>> I hope you make the comments above on the blog.
>>>
>>>
>>>
>>> Adrian
>>>
>>>
>>>
>>>
>>>
>>> --
>>>
>>> Adrian Gropper MD
>>> Ensure Health Information Privacy. Support Patient Privacy Rights.
>>> http://patientprivacyrights.org/donate-2/
>>>
>>>
>>>
>>
>>
>>
>> --
>> Adrian Gropper MD
>> Ensure Health Information Privacy. Support Patient Privacy Rights.
>> *http://patientprivacyrights.org/donate-2/*
>> <http://patientprivacyrights.org/donate-2/>
>>
>>
>> _______________________________________________
>> Openid-specs-heart mailing list
>> Openid-specs-heart at lists.openid.net
>> http://lists.openid.net/mailman/listinfo/openid-specs-heart
>>
>>
> _______________________________________________
> Openid-specs-heart mailing list
> Openid-specs-heart at lists.openid.net
> http://lists.openid.net/mailman/listinfo/openid-specs-heart
>
>
>

-- 
Adrian Gropper MD
Ensure Health Information Privacy. Support Patient Privacy Rights.
*http://patientprivacyrights.org/donate-2/*
<http://patientprivacyrights.org/donate-2/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-heart/attachments/20150420/177d5495/attachment-0001.html>