News for hackers. Stuff that matters.

maddog99

Re:How a robot pharmacist wound up giving an overdose on Thursday April 02, 2015 @04:42AM (#9309348)
My proposal: usability experts should identify an acceptable range of alerts on a per-order, per-encounter basis. End-users of all disciplines should have the opportunity to provide feedback in real-time, in the live environment. Just like Amazon reviews, "Was this alert helpful?" If you always vote "no", your feedback will be down-weighted. With a few exceptions, like wrong patient or 10x overdose, all the other alerts will be ranked in order of utility and only the top x% will be shown.
Re:How a robot pharmacist wound up giving an overdose on Thursday April 02, 2015 @04:29AM (#9309291)
As a sanity check, what about showing the ordering physician a picture of 38.5 tablets? "Is this correct?" I think that would grab my attention.
Re:How a robot pharmacist wound up giving an overdose on Thursday April 02, 2015 @04:25AM (#9309277)
Not sure what "intimacy of care" means, but most docs, like me, are frustrated and annoyed by the barrage of wolf-crying that comes out of these systems, of which Epic is one of the best and best-designed. (So far, all the others I've seen are FAR worse.) I WANT to see the meaningful alerts, but it's difficult to do with a sea of identical-appearing utter BS. Besides alert fatigue, there's a related problem with cognitive load. For this patient, I want to order A, B, C, and D. Now erect roadblocks in front of A, B, and C, and quite often D will fall by the wayside. By the way, I've never been able to set or even influence the alert policy. This was mandated by corporate risk management, whose policy appears always to be more alerts = safer.