I'd be interested to hear from people that were around when expert systems were in vogue the first time around as to why they didn't take off in medicine, and what is new (if anything) in the new wave?
I think we are getting there. It's not easy to build an expressive and efficient formalism that combines structure and uncertainty. New approaches like Markov logic are promising.
Why not "Your New Medical Team: Computer Scientists and Physicians" ...
Probably broader than just CS -- wouldn't be surprised if the teams include physicists, engineers, computational scientists, etc.
One challenge ahead--in most hospitals, the doctors have to choose to use these new tools and many aren't. Some don't see the need (maybe due to a generally conservative attitude), and others actually see them as a liability (if there's more data than they can personally review, they could be sued for missing an important signal). But at least the research hospitals are making headway in developing and testing new tech.
Right on point. I work for a firm that helps predict whether patients will be readmitted for various reasons, and while our algorithms are better than any standard rules, getting doctors and case managers to actually use and trust them is the greatest challenge we face.