I’m skeptical. Do you have an argument to back up that claim? Doctors and nurses that provide direct healthcare are more than 1% of workers, are you saying AGI that replaces a doctor can’t be considered “strong AI”?

A doctor performs a well defined task: he collects data about a patient as inputs to his mental model, and produces diagnosis as the output (and selects a treatment based on established guidelines for the diagnosis, and relevant patient characteristics). I don't see anything in this process that requires AGI. In fact I think a doctor is one of the easier professions to automate, and it will probably be automated relatively soon.

Nurses which provide emotional support to patients might be harder to replace with AI, but that too will change as AI starts to become more and more "human-like" (i.e. more complex), but not necessarily AGI level.

Speaking as someone who works in the space, this is wildly optimistic. All I can say is your view of AI is pretty off. You need pretty fantastic AI just to understand what a patient is telling a doctor, let alone make a diagnosis.

Looking at national stats for medical errors, and given how bad most doctors I interacted with are at listening, and making diagnosis, I'd say the bar for AI is set pretty low.