I was at Tufts University and I happened to get connected
with Thomas Trikalinos and Joseph Lau, and Chris Schmid.
And at the time I was working on machine learning and
natural language processing stuff.
And it occurred to all of us jointly, along with my advisor, Carla Brodley,
that perhaps we could expedite some of the processes that were being done manually
in evidence synthesis using machine learning and natural image processing.
So ever since then I've been working on that.
The aspect that's received the most attention is probably on
the screening side.
So this is where you're identifying the literature,
the evidence to be included in the systematic review that you're performing.
And there's been a lot of work that's looked at using automation techniques, and
in particular, automatic classification techniques to expedite that process.
And I think those technologies are actually pretty well established.
And I think the general consensus is that you can reduce the labor by about half
without sacrificing sensitivity or recall to the relevant evidence,
which is really important.