The front-door decision: where AI could help patients choose the right care

The most consequential triage decision in healthcare is made before a patient ever meets a nurse. It happens on a couch at 11 p.m., when someone decides whether a symptom means the ER, urgent care, or waiting until morning.

As an emergency physician, I see the results of that decision every shift. Some patients come in for problems urgent care could have handled. A few wait at home with problems that could not wait. Both cost the system, and the second can cost a life.

Why this is an AI question

Patients are already asking chatbots this question. That makes the front door the place where patient-facing AI has to be most careful. A tool that is right most of the time but misses a stroke is not a good tool.

I build every patient tool on my DR GPT™ Labs sites around three rules:

1. Emergencies first. The first thing on the screen is when to call 911, before any other answer.
2. Examples, not diagnosis. The tool helps a person sort symptoms into "911, ER today, or urgent care." It never names a disease.
3. Nothing leaves the browser. No account, no stored symptoms, no data to protect.

What leaders should take from this

If your system is putting a symptom checker or chatbot at the front door, ask three questions before launch: What does it say first when someone describes chest pain? Who reviewed the wording clinically? And what data does it keep? Those questions matter more than the model behind it.

You can try my free ER or urgent care helper to see the approach. It is the same thinking I bring to boards in my keynote on SAFE AI.

I bring this thinking to hospital boards and conference stages. Book me to speak.