Spaces:

---

Agents-MCP-Hackathon

/

HealthMateAI-Medical-Triage-Assistant

like 16

Running

Hi, pretty nice functionality!
I would add more inputs like scales for: "pain levels", "hinders me in daily life". Mainly, that the user thinks about the urgency himself :)

You could also output a script which helps the user explain the situation to the receptionist and doctor. Or just some keypoints to use in the phone call

I really like your suggestion about adding input scales like “pain level” and “hinders daily life.” It’s a great way to help users assess urgency themselves. I’ll definitely look into incorporating that.

Also, your idea about providing a script or key points for users to explain their condition to a receptionist or doctor is excellent. That would really improve usability, especially in stressful situations.

Really appreciate your input it’s very helpful as I continue to improve HealthMate.

I'm a doctor in my professional life. Feel free to let me know if you want to brainstorm on this tool together.

I have to admit it works pretty well, but I found it can make mistakes when the symptoms are worded in a slightly different way.

For example:
"I have fever with mild shaking, feel slightly less alert, sweaty skin, and pain." --> it correctly diagnoses this as a possible sepsis and therefore labels it an emergency that requires urgent attention (which it does).
"I have fever with mild shaking, feel a little bit hazy, sweaty skin, and pain." --> it labels it as a routine visit within 24-48 hours or so, during which a developing sepsis can definitely become lethal if left untreated

I think the concept is great, but I feel a disclaimer should be put in place. (i.e. to avoid: "the AI said it wasn't that big of a deal so I thought I'd just tough it out for a bit until I suddenly got a lot worse").

Cheers and good luck with your impressive project!

*Edit: it probably triggers on the word "alertness" since that is how it's commonly described, but as demonstrated with this quick example here that is not necessarily how a patient might word it to you.

Another thing that's missing that I feel is vital is the option for a patient to input their medical history and their list of current medication.
These can make a world of difference whether or not a symptom is "not that big of a red flag" or "holy shit call the entire trauma team into the room".

Just a few random (and not the best) examples:

1. a slowly developing headache without much other symptoms could be not that alarming, but a slowly developing headache 2-3 days after an elderly lady on Apixaban hit her head against the dresser drawer might be very serious.
2. A little shortness of breath might be not that alarming, but in a pregnant lady it might indicate a pulmonary embolism. Or what about that guy that just had that 24 hour flight last week?
3. Vomiting and diarrhea for a day is probably fine, unless you take external cortisol and are about to dive head-first into an Addison crisis without immediate intervention because you can't keep your medication down long enough for it to be absorbed.

(I realize that technically you allow for that information to be entered, but it's not really inviting to do so in it's current state so a lot of people might not bother. You could argue that doing that takes too much time, but lets be realistic, the majority of people that will be using this are not going to be dead within the hour. Even so, you could make it easier by for example providing some major groups of pharmaceuticals that people can just click on quickly.).

HI CultriX,
Just letting you know that your opening line about being a doctor in your professional life, Made my Day. That I'm receiving inputs and suggestions from the actual professional. My Goal with this tool is to make it people's helping hand not their only hand so they over-rely on it causing some serious issue(I hope it would never happen) So I will make sure to put a disclaimer asap.

if you're open to it, I would love to connect with you to discuss and brainstorm this tool further updates and maybe even walk you through the updated version before releasing it again. I've noted down the things you said(genuinely Thanks) your feedback has already improved the direction of this project significantly, and I'd be grateful to continue learning from your perspective. Let me know which platform you're comfortable using to discuss this further.

Thanks again for your kind words and for pushing the project to be safer and more clinically aware.

Hi, thanks for the kind words! :)

Would Discord be ok with you? cultrix_14764