Roupen Odabashian, Oncologist at Abbotsford Regional Hospital and Cancer Centre, Founder at MeDucation AI, Podcast Host at OncoDaily, shared on LinkedIn:
“The biggest real world study of AI medical scribes came back with an uncomfortable answer: modest time savings, and highly inconsistent use.
This is the technology everyone pointed to as the obvious win. Ambient AI listens to the visit, writes the note, gives the doctor their evenings back. Epic, athenahealth, and the VA are all rolling it out at national scale. A UCLA study even found documentation time dropped by roughly 10 percent for users.
Ten percent is real. It is also a lot less than the story we were told, and the “inconsistent use” finding is the tell. A tool only helps if clinicians actually trust it enough to lean on it in a busy clinic.
The lesson is not that AI scribes failed. It is that deployment is the hard part, and adoption is a clinical behavior problem long before it is a model problem.
We keep measuring these tools by benchmark scores. What would change if we measured them by whether tired clinicians still use them at 4pm on a Friday?
Research on AI scribes and CDS tools are becoming useless! you can tell any story by choosing the metric that you like!”
Other articles about AI in oncology on OncoDaily.