Embed code
a one question dangers i hope i'm not a watch that i run on a light
thinks he should that people might be called a lot of i know we can go
um we haven't addressed that at all but it it is definitely a problem in there and like car
i think there there are many dependencies that we haven't looked
into i think the exact complication people like in
in in our system definitely um optimise it on the available infrastructure
and clout so they knew like what infrastructure they had
things we were running in the document anus who if we can make
like we can like work flow soft taco containers so not have
like one big monolithic system but having like for every test that's what
we did like four then a quantitative imaging that work well
a system there's like some tasks so at all the different groups that meet docket containers
for the summer tossed and then we creative work flow we can combine them
to actually run it on a lot of times he automatically we have like some
care relies ability in in the system that we have there's also problem with like security
so if we move code inside a hospital like you you can half
that call in utah container so we may need to make sure
you cannot proceed great like a fully sandbox environment will be run it
and like totally disconnected from the rest of the hospital network
and then delete the virtual machine in the end and only like extract the results so there
are ways of dealing with that but it's something to to take into comes on
yeah well
no one else actually yes
any other question is to have a one
right and also yeah one action um examples of a project level
internal visible sorry medical research just so they'll for it
that computers are put the algorithms that would be evaluated on yeah they told
me that she was but my question with this kind of models
what's the price like so uh because putting the shoes that you'll cooled on so that they don't the company
and we can run sings what you live so doesn't mean that you
give them some kind of uh uh lights on the software
before they all you see the the you know the and it's it's it's totally true
like in in the case of intel they participated in in one of our workshops
and they were rather presenting like their vision for the medical domain of
easy for them hospitals apply it's potentially because with all electronic analysis
now for some like in the massachusetts general hospital were working boston um
they do systematically every cancer patient gets the whole genome sequence
and i'm like that they can be that therapies to it so if we do that
like we need massive computing infrastructure so for them to have the infrastructure
inside the hospital it doesn't mean that they run the infrastructure
they are they're not responsible they just loving heart pretty much so in that respect
it's not a problem because it's the hospital wanting to code but there is obviously
uh if if i'm a company into run the code in someone else's system
there could be like reverse engineering the could be like our analysis up the code even if we don't have source code to be
so it is like all we get is executable so people don't need to you to submit source code but we had companies
in general electric wanted to participate in the desert competitions and they ask us um
like can you guarantee us that nobody will touch a coding nobody would be able to see a coat
i'm not a security specials i told him i can't and in the end they didn't participate
because they said like oh no we need this guarantee to have that so i'm
i think a lot of it is about trust 'em between partners but
the result the seals the problem off are like what we can
do for the silver corpus we asked the participants whether we can use the code to run it on your data for example
alright thank you lose one people the next year i think

Share this talk: 


Conference program

Welcome
Sébastien Marcel, Senior Researcher, IDIAP, Director of the Swiss Center for Biometrics Research and Testing
24 March 2017 · 9:17 a.m.
Keynote - Reproducibility and Open Science @ EPFL
Pierre Vandergheynst, EPFL VP for Education
24 March 2017 · 9:20 a.m.
Q&A: Keynote - Reproducibility and Open Science @ EPFL
Pierre Vandergheynst, EPFL VP for Education
24 March 2017 · 9:54 a.m.
VISCERAL and Evaluation-as-a-Service
Henning Müller, prof. HES-SO Valais-Wallis (unité e-health)
24 March 2017 · 11:35 a.m.
Q&A - VISCERAL and Evaluation-as-a-Service
Henning Müller, prof. HES-SO Valais-Wallis (unité e-health)
24 March 2017 · 12:07 p.m.