Context: I’m a second year medical student and currently residing in the deepest pit in the valley of the Dunning-Kruger graph, but am still constantly frustrated and infuriated with the push for introducing AI for quasi-self-diagnosis and loosening restrictions on inadequately educated providers like NP’s from the for-profit “schools”.
So, anyone else in a similar spot where you think you’re kinda dumb, but you know you’re still smarter than robots and people at the peak of the Dunning-Kruger graph in your field?
At least I can rest assured of the fact that AI will be next to useless in my intended field. Emergency medicine is an environment where you get a random constellation of symptoms and complaints with very little direction on which are related to the current illness, and which ones are not currently relevant. Also, in the time it would take to get all the info into the AI for a trauma/cardiac/code situation, the patient might be dead or rapidly heading in that direction.
Can’t AI aggregate the data on triage outcomes to prioritize larger scale emergencies… it has to be useful somehow I would think
It wouldn’t need to be AI, just sone statistics and “Crdiac arrest? Piority 9. Broken arm? piority 1” decision-making.
I’m a tech wizard not a healer, there probably are factors that define one cardiac arrest even more critical than the other i just do not know them.
There’s some things you look for that are difficult to describe to someone who hasn’t seen it before. That’s part of why experience is so valuable in Emergency Medicine, and it’s not uncommon to put your best nurses out in triage. People will do this kinda twitchy/wilting/loss of focus/change in pallor/change in posture right before they go down. I don’t have a good way to describe it, and it might be easier to draw even, because it really is a body language thing and the general appearance of the patient that can inform your decision making.
I have thought about trying to plan out a learning algorithm that could spit out suggestions for triage level and preliminary tests based on input data like vital signs, symptoms, and complaints… but I would never implement something like that as anything beyond a tool for the nurses at triage to use. There would have to always be an option to override the algorithm because there’s some aspects of patient presentation that are not easily quantifiable. I’d never be able to explain it in a way that one could input it into a computer, but even with my limited experience, I can tell which patients are going to crump on me.