How AI Helped Me in the ER

I woke up on Thursday morning with such severe abdominal pain that I couldn’t even get out of bed.  My husband was in the shower, but I was unable to stand up to ask him the telltale question: “Do I need to go to the hospital?”

It’s always such an extreme question. We have two kids, it was 7am, the start of a workday before the start of the official school year, which meant the baton passes of kid duty had been meticulously schemed weeks in advance. One parent’s absence would wreak total havoc. So if one parent thought they might need to go to the ER, it was important to be really sure. 

Remembering back to my earlier blog post, “Could AI help in my next at-home crisis?” I decided to do the things I said I would and turn first to ChatGPT. I’m not joking when I say that I was in so much pain at that moment that it hurt to even type. I wrote:

Really painful abdominal cramping 
All sides 
Mostly in front  
Hard to walk 
Also feel pressure in pelvic area 

Upon waking up 
Drank a lot 
maybe hungover 

Feeling hot and sweaty 
Ate beer tartare last night 
Maybe food related?

My initial response indexed a bit too strongly on the eating and drinking:

It continued to suggest a few other muscle concerns, dehydration concerns, and general advice. But I had a feeling this was not stomach pain, so I doubled down on the acute nature of the abdominal pressure, indicating that the pain was so much that I could not walk.

From there I received a shorter list of options:

That’s when I decided to go to the Emergency Room. 

Several hours later, I found myself explaining my process to the attending physician.

By this point, the doctor had deduced that an ovarian cyst had likely ruptured earlier that morning, causing the severe pain and pressure I’d initially experienced. He said it was good that I’d come in. (Disclaimer: I'm fine now, and spent the weekend at home, in recovery, on much medicine.)

I confessed that I’d gone first to ChatGPT to help me decide. 

“Oh yeah?” he said. “What did ChatGPT suggest?” 

I had my laptop in front of me so I just pulled out the chat log and showed him the initial read, and then my subsequent inquiries that led me to deduce it was more of a pelvic issue than a stomach issue, and what eventually led me to come in.

He looked thoughtfully at the screen, then back at me.

“Wow,” he said. “The AI is nicer than me. It started off by acknowledging your discomfort and pain, I didn’t even do that.”

I glanced back at the very first message readout and sure enough, right at the top, before we got into any symptom triage, it said:

Huh. I thought. Isn’t that interesting?

In the 8 hours I spent in the ER, I spent 7 of them completely alone.

My husband had to manage the kids all day. The doctors visited only sporadically. On multiple occasions, I found myself walking through the crowded aisles in a hospital gown and socks, asking for basic things—food, water, pain medication, a blanket.

The backlog in the ER was so severe that I was receiving text updates on my lab results before the doctors had time to come in and explain them. Before the doctors even arrived to walk me through the findings, I’d run the results through ChatGPT, texted my OB, and began drawing my own conclusions. 

Is this really the best way for a system like this to function?

One of the biggest fears that I hear from folks about emerging new technologies like AI is the worry that it will replace our jobs. I get that. I really do.

And yet. I also wonder: Can this technology actually buy us time back to spend doing the things that only humans can uniquely do? Things like, spend a little more time with each patient in the ER, as a human? 

The more technology I adopt into my workflow, the more I am able to accomplish. And furthermore, the more time it frees up for me to have space in my own day to do some of the smaller things. Time to chit-chat with my neighbors. Time to write another blog post. Time to read another book with my kids.

The doctor’s first reaction to the AI’s readout wasn’t to fact-check his own diagnosis—it was to acknowledge that the AI had recognized my discomfort. That was telling. It made me realize that doctors know they need more time to be human with their patients. 

Maybe we’re looking at it from the wrong angle: AI isn’t taking over our jobs; it’s freeing us to focus on what humans do best.