open source medical software

[Donnal Walter]

--- Kelly Anderson <kelly at acoin.com> wrote:

> It's probably obvious from the previous replies, but I think
> it's worth stating that you might consider breaking the
> project up into several pieces, ...

We have separated the development framework from the applications.
This framework is clearly standalone and GPL or OSL would be
perfectly appropriate. We also have in mind developing a suite of
clinical applications using this framework and then evenutally
combining these components into an integrated system.

> ... each of which has it's own license. This can give you
> more flexibility than one license might be able to. You
> could get the benefit of open source for the parts that
> are perhaps more generally applicable (many eyes and so
> forth) and largely avoid the problems with the more 
> specifically medical parts for the majority of users. This
> would increase the number of users for the parts that are
> more generally applicable, giving you more of the benefits
> of open source to your overall project.

Yes, this idea would work with the development framework, but I am
also interested (1) in making the clinical applications freely
available, and (2) in obtaining feedback and improvements from
clinicians. I don't know that FDA regulations preclude this; I am
merely trying to get the lay of the land. 
 
> I'm curious about something though. As I understand it, the
> FDA tends to look at "medical devices" as a whole system,
> hardware, software and processes together.

In general, I think you are correct in this, but in general, a
medical device has some fairly direct access to the patient.

> Does that mean that a Computer running Linux has to be 
> FDA approved prior to running an X-Ray machine with it, or
> examining X-Rays, or whatever.

The reason I raised the questions I did originally is that I don't
have all the answers. That said, I am pretty sure that operating
systems and database managements systems need not be approved as
such by the FDA.

> The point here is that if the FDA has approved "computers" and
> specifically "Linux" or "Windows XP", then there is precedence
> for how they might deal with a multi-part piece of software. 
> Similarly, they may have approved the use of Oracle in a system,
> but maybe not MySQL, or maybe they have approved both. If the
> "base" package can get initial FDA approval, then it's only the
> smaller part (that might change more frequently) that would have
> to go through the process regularly. I guess my point here is to
> find out what the FDA has done with other open source projects
> that may have been incorporated into medical devices, because
> there had to have been one or two previous attempts. This also
> goes to the idea of separating the pieces, as a general "Baysian
> Engine" might get FDA approval on it's own more easily than the
> entire system.

As I said above, I don't think the FDA looks at general purpose
software, even when it is used in medical computing. I don't know
that for a fact, however, and I intend to find out. It is a good
suggestion to find out what the FDA has done with other open source
projects, but up to this point I have not seen this mentioned,
either in the documentation of open source or in any of the FDA
documents I have looked at. I am not saying the information is not
there, but a few quick Googling sessions turned up nothing for me.

The most useful information I could find is this:
http://www.fda.gov/cdrh/comp/guidance/938.html
and an older article at:
http://www.devicelink.com/mddi/archive/97/10/019.html

Ultimately we should probably ask the FDA our specific questions,
but before we do, we want to make sure we know how to ask the right
questions.

One thing we have in our favor, I think, is that none of our
software will interface directly with what the FDA considers a
"medical device" either on the input side or output. Moreover, the
users of this software are practitioners who make independent
medical decisions on a routine basis all the time. We are merely
trying to make this process more efficient, not change the
fundamental nature of the decision making. YMMV.

Regards,
Donnal Walter