Although sometimes overhyped, big data technologies
do have great potential in the domain of computational
biomedicine, but their development should take place in combination
with other modeling strategies, and not in competition.
This will minimize the risk of research investments, and will
ensure a constant improvement of in silico medicine, favoring
its clinical adoption.
We have described five major problems that we believe need
to be tackled in order to have an effective integration of big data
analytics and VPH modeling in healthcare. For some of these
problems there is already an intense on-going research activity,
which is comforting.
For many years the high-performance computing world was
afflicted by a one-size-fits-all mentality that prevented many
research domains from fully exploiting the potential of these
technologies; more recently the promotion of centres of excellence,
etc., targeting specific application domains, demonstrates
that the original strategy was a mistake, and that technological
research must be conducted at least in part in the context of each
application domain.
It is very important that the big data research community
does not repeat the same mistake. While there is clearly an important
research space examining the fundamental methods and
technologies for big data analytics, it is vital to acknowledge
that it is also necessary to fund domain-targeted research that
allows specialized solutions to be developed for specific applications.
Healthcare, in general, and computational biomedicine,
in particular, seems a natural candidate for this.