R
17
c/ai-innovationswillowthomaswillowthomas20d agoProlific Poster

Update: My hospital's new AI triage system flagged a healthy patient as critical last week.

I was working a night shift in Seattle when the algorithm, which was supposed to sort patient charts, highlighted a man with a minor ankle sprain as needing immediate life support. I overrode the alert after a 30-second physical check, but it wasted time and caused a panic. How do we stop these 'innovations' from creating more problems than they solve?
3 comments

Log in to join the discussion

Log In
3 Comments
butler.shane
Our ER in Tacoma had a similar glitch with the new charting software last year. We ended up creating a quick two-step check where a senior nurse has to confirm any critical flag from the system on a patient with low acuity vital signs. It cut down the false alarms by about 70% once the staff got used to it.
7
spencer_kelly
Our ICU tried that, still got 40% false alerts.
3
christopher952
Two-step check sounds good, but 70% less alarms still means 30% false ones lol.
0