AI better in the ER room

If your image of an ER room is of complex, crazy, life threatening and changing decisions happening in the moment, you are not wrong.

Could AI cope with this?

The answer is yes. Harvard researchers found an AI model (LLM) could conduct real world triage in emergency rooms, recommend appropriate diagnostic tests, and perform case management tasks at a level that matched or surpassed the ability of even well-trained human doctors.

They tasked the LLM to arrive at a patient diagnosis and develop a testing plan, then evaluated its skill in clinical reasoning compared to both experts and generalist physicians.

In a live clinical setting, they assessed the LLM’s performance on 76 emergency room cases in a Boston hospital at three stages: initial triage at arrival, first contact with a physician, and upon admission to the medical floor or intensive care unit.

The AI was particularly good at making assessments at the initial triage stage, when there was the least information available and at diagnoses involving rare diseases and complex cases.

There are some limtations. The study was based entirely on text-based inputs. Practicing physicians evaluate many other forms of information from listening to the patient, reviewing imaging studies, and physiological signals like EKGs.

The researchers noted the improvements with the latest models. Does this replace doctors, the researchers are clear in saying No.

But it does provide easy access to a second pair of eyes, or to help form their opinions in complex, fast moving environments and there are some things the AI does better than humans. The better way to think about it is as a tool to help physicians do their work.

SOURCE

https://www.harvardmagazine.com/ai/ai-outperforms-doctors-diagnosis-harvard-study