Don’t just automatically trust a diagnosis from ChatGPT please. Consider it as something that will cast a wide net covering every possible thing you ‘could have’ at a high level. You will need to provide it a LOT more info and keep refining your prompts to reduce that net.
I tried ChatGPT for very odd symptoms I have and it was way off from the actual diagnosis and I’ll explain a little my experience.
I’m a data person so I created a ChatGPT project and loaded the following:
1) Individual MRI’s, CTs, X-rays for past 3 years
2) All clinical doctor notes as individual files from past 3 years
3) A download of all bloodwork from past 2 year formatted as a consolidated table (had chatGPT do that for me)
4) A table of all active medications
5) A download of all medications filled over past 2 years from my pharmacy website
6) A detailed list of all my current supplements including brand name, dosages, and an Amazon link to the exact product
7) A detailed list with description of all my symptoms including: frequency of them, start of the symptom severity level (severe, very high, high, moderate, mild), presentation (constant, cyclical meaning if only when x occurs or only in mornings, sporadical) description of the presentation (e.g. pain in lower back is located on the right side and is ‘severe’ in the morning waking me up), what makes my symptom worse, what makes my symptom better (e.g. Pain reduces down to ‘moderate’ or ‘mild’ after my morning walk), symptom sensations (e.g. pain is deep and burning or pain is sharp and sudden), other symptoms that may also occur when this symptom is worse
8) List of all therapies I have tried, dates, frequency and duration (e,g. 30 min, 3x week), was the therapy effective, if therapy is not active reason for stopping
9) list of specialists I have seen, dates, if they are actively treating me, if I am not seeing them why, summary of what diagnosis or suspicion of diagnosis I seeing them for
10) list of past surgeries ever in my life
11) list of serious injuries or accidents ever in my life
12) family history

I then start ofd with a simple first prompt and say “if you were an internal medicine doctor, what questions would you want to ask this patient”. Then answer all those questions and ask ChatGPT to commit these answers to memory as an artifact for your Project (not just your chat)

Next prompt “as an internal medicine doctor, what specialist referrals would you recommend for this patient and why. What order should they see which specialists”

Next prompt “if you are the first recommended specialist, what questions would you have for this patient”. Answer and ask it to commit your answers to memory as in the above prompt. Keep asking this prompt until you have gone through all recommended specialists.

Next prompt “update my symptom list based on my answers provided so far and generate in a printable table view while committing this as an artifact to my Project (not just the chat)

Next prompt “analyze my uploaded bloodwork, list of medications and supplements and updated list of symptoms. Provide any insights into any medications or symptoms that could be contributing to my symptoms or bloodwork levels. Explain as if you’re talking to a general patient and use easy to understand terms including what the impact could be, why does it cause an impact, and are there any alternatives I should discuss with my doctors.”

Next prompt “do you see anything in my list of past injuries and accidents or surgeries that could be impacting the symptoms I have today”. Continue a q&a with it till you understand as scar tissue (say from a hysterectomy can cause some back, nerve or muscle issues as in my case). Important- ask it commit a summary of the impacts to it’s memory and as an artifact for your project.

Note I NOT asking for it to diagnose. Next prompt “I am the patient, put an easy to understand list of things I can start doing at home to ease my symptoms while I wait to see the doctors. This list should be generated using an internal medicine practitioner view and so not included anything they would not typically suggest”

Next prompt “using everything you know so far what other questions do you think an internal medicine doctor or specialist would have for me that have not already been answered”. Answer them and ask it commit to its memory and as an artifact in your project

Next prompt “prepare a printable packet for each doctor I can take with me to my appointment and include: a page of numbered questions I should ask, a reference sheet for me of reasons for asking the questions that are number to match the questions, exhibit pages for the doctors using sections from my artifacts that the doctor would be interested in knowing about - make the exhibit part of the packet numbered and pull that number onto my reference sheet so I can also understand how they may relate to my questions”

It is also a good idea to always ask ChatGPT to re-review everything in the packet again against all artifact in the project and ask it what else it should include that it may not have, or ask it to think again and again if there are any other questions it feels should be answers by you to ensure you have the most productive discussion with your doctors.

Please don’t gaslight yourself by asking it what it thinks is wrong or what the diagnosis is. If it tries to give you that information steer it back to information for the doctor.

The sequencing of the prompts is key so it ends up having all the information it needs to ensure you have enough information for your doctors as they are the only ones who can diagnose. Doctors can’t spend hours with you so the more information you have the better as it will help you get to a diagnosis quicker and reduce your chances of a misdiagnosis. Remember to be a partner and collaborator with your doctor as we are all complex beings and symptoms could occur with 100’s of different things. Also, just because you have spinal degradation doesn’t mean something is ‘wrong’ it could be part of aging.