afib and ChatGPT : just to say - I’ve... - Atrial Fibrillati...

Atrial Fibrillation Support

33,536 members39,828 posts

afib and ChatGPT

Jimbobarelli profile image
43 Replies

just to say - I’ve found ChatGPT an absolute fountain of useful knowledge - everyone’s AFIB is different and this AI tool can even access your iPhone health data to give you tips on what’s going on with your heart, the drugs you are using etc .

Written by
Jimbobarelli profile image
Jimbobarelli
To view profiles and participate in discussions please or .
43 Replies
ozziebob profile image
ozziebob

Yes, I'm sure there will be positives, but please also be aware of potential errors and harms. The persuasively seductive language used in answers should itself raise alarm bells. My experience is mixed. I was given a very impressive list of research articles (and links) related to a health area I was investigating (polio and mitochondria), but I discovered most of the articles were either unavailable, required payment, or no longer existed. Very disappointing!

Just be careful out there, and do research from other sources to verify the ChatGPT answers.

Jimbobarelli profile image
Jimbobarelli in reply toozziebob

yep. I do check. But what I’ve found is that I have various questions pop up in my head at regular intervals( daily!) - recently for example I had moved to taking bisoprolol for my PAF and I’d noticed that when I was doing interval training , my heart seemed to take a few rounds before it got into a normal ‘ active’ and ‘ recovery’ rate . I actually put the heart rate history data file of the session into chatGPt and it explained what was going on and why everything looked ok. The other thing is that it gives you different views on the validity of some drugs over another - which you don’t get sometimes from your cardiologist who may have his set views on treatment . It’s allowed me to better understand what’s going on with my episodes and the many things you can try to minimise them .

ozziebob profile image
ozziebob in reply toJimbobarelli

I only have the free version of ChatGPT, and I understand the paid for versions have more impressive options available. But I won't be paying.

As for entering data, sadly I haven't mastered that yet, so well done you. I suspect you are are a younger generation than me. Another member did mention entering her echocardiogram results and was rewarded with helpful comments. So I'm keeping an open mind, as best I can, for a 78 year old fixed air sign.

Ppiman profile image
Ppiman in reply toJimbobarelli

Who needs doctors, eh? 😉

Steve

Jimbobarelli profile image
Jimbobarelli in reply toPpiman

That’s not what I said ….

Ppiman profile image
Ppiman in reply toJimbobarelli

I know, hence the 😉. I was joking. Sorry if it came over otherwise. I can imagine the NHS moving to AI doctors at some point.

It was also because you said cardiologists didn’t necessarily prescribe effectively, too. They have their favourites based on their learning and experience, I guess.

Steve

Jimbobarelli profile image
Jimbobarelli in reply toPpiman

It’s just another tool to use . I’ve down a lot of sport with quite a lot of self awareness on my heart , pulse , recovery etc so I’m not blindly following AI but rather letting it give me a few ideas I may not have considered . It’s cheaper than a private doc , and more easily accessible than my NHS GP !

mjames1 profile image
mjames1

AI it's a great tool, but still at the point where it's making a lot of mistakes. It definitely sounds brilliant in subjects I don't know a lot about, however, on subjects that I am well-versed in, I often pick up mistakes.

Definitely cross check anything on AI with google for credible references, that you may be basing a decision on.

As for inputting a lot of test data, just be warned that it's programmed to always sound intelligent and authoritative, but again it makes mistakes and you would never know it unless you're an expert on the subject or you cross check.

One of the best doctors I ever went to at times iresponded to complex questions with "I just don't know or I'm really not sure". The chat box will never do that, but that doesn't make it smarter than that doctor. To the contrary.

Jim

timetabler profile image
timetabler in reply tomjames1

You wrote "It definitely sounds brilliant in subjects I don't know a lot about, however, on subjects that I am well-versed in, I often pick up mistakes."

How true! I find the same thing.

mjames1 profile image
mjames1 in reply totimetabler

Pretty much like the news media!

timetabler profile image
timetabler in reply tomjames1

But worse ...as AI can sound so personal and authoritative.

Responsable profile image
Responsable in reply totimetabler

Yes, just another means of manipulation, aimed for control and brainwash...

ForensicFairy profile image
ForensicFairy

I’m with you on that. It’s far better than Google and gives detail in the way you want it receive it. I find that those who think negatively of it haven’t used it enough to understand how it differs from googling. It’s a different concept but people tend to try and use it in the same way with varying results.

I have fed mine all my cardiac info. It knows my conditions, drugs, and the way I like to receive information. It holds memory of certain points and you can direct it to include or not include information. So the person who has the search result filled with links that weren’t accessible - you simply tell it to ensure all links in future are accessible, peer reviewed and relevant. It’s that easy.

I had issues with mine giving me American based information and spelling. I asked it to only suggest local things when appropriate and to always use the ‘queen’s English’ for spelling and grammar.

I use it even for simple things in life - I might have a couple of items in the fridge and can’t think of what to cook so I tell it what I have in the pantry and fridge and ask for a recipe and it gives it to me! Simple. The recipes have been great.

Recently I asked if for some stretch band exercises to help me keep muscle tone. It did. You can ask it to refine or focus the exercise for whatever purpose specifically and it will. Have a bad knee - tell it and it will ensure the exercises suit your knee too!

Elli86 profile image
Elli86

hi Jim. Do you have the new iPhone with gpt integration then? If not how does it gain access to your health data?

Jimbobarelli profile image
Jimbobarelli in reply toElli86

You can “export all health data “( click on your icon top right of the health app ) as a zip file - or individually copy and paste specific data or charts

mjames1 profile image
mjames1 in reply toElli86

No. Older phone.

fairgo45 profile image
fairgo45

I absolutely love chat since I discovered i only have to screenshot my echo results or wellue results tap on + photo down bottom left of screen and press the arrow to upload then tap analyse. I say I'm.in New Zealand it's so much clearer than trying to understand the cardiologist medical jargon and easier to compare notes

Chat is always very empathetic .

You can interact easily with it and it will remember all your previous results if you haven't tried it I throughly recommend you try it

ozziebob profile image
ozziebob in reply tofairgo45

I remember when you previously mentioned uploading your echo results. Now you have explained the process in more detail, I explored further, but my free version ChatGPT wanted me to create an account before allowing access to the features you mentioned. As I have previously, by choice, tried to use ChatGPT as anonymously as possible, I obviously have doubts about the wisdom of loading any personal data into another potentially dangerous online data base. It seems I will have to survive without the "benefits" of creating a ChatGPT account, as detailed above by ForensicFairy & Jimbobarelli .

fairgo45 profile image
fairgo45 in reply toozziebob

I don't think think there's any thing to worry about if you want to be anonymous when up loading anything anyway don't give your personal details I always crop out any addresses or names before uploading it.if your worried just say forget everything we've talked anout and all you've shared will be forgotten.

ozziebob profile image
ozziebob in reply tofairgo45

Do you have a ChatGPT account? Are you using the free version?

fairgo45 profile image
fairgo45 in reply toozziebob

Just the free version it limits the amount of times you use it but refreshes the next day so you can use it again.

ozziebob profile image
ozziebob in reply tofairgo45

But do you have an account?

fairgo45 profile image
fairgo45 in reply toozziebob

Yes I have an account not using my real email address.

ozziebob profile image
ozziebob in reply tofairgo45

But that's what I said in my initial reply to you. I couldn't upload a photo as you did with your echo results because I didn't have an account and you need an account for those options.

Are you saying you used an email account that doesn't exist to create an account, or something else? PM if you would rather not reveal details.

ForensicFairy profile image
ForensicFairy in reply toozziebob

I only had to create an account with an email address. I also have a free version and can upload photos.

Another handy thing is that it will also create pdf and word docs for you - as an example, I was seeking info on a new drug I had been started and it recommended regular BP measurements etc. It created a table for me to use to record this and asked if I wanted it as a pdf or Word version. Simple, but it was handy and I didn’t have to create it.

This is also another handy thing it has - although you could also search your history for Google results I guess. This is a snapshot of previous questions I’ve asked it. It’s very handy to refer back to things. I’ll also try to find an interaction I had with it when I challenged it about COVID vaccinations.

Just like Google, people need to use their critical thinking skills and question things.

Screenshot showing list of ChatGPT engagements.
fairgo45 profile image
fairgo45 in reply toozziebob

I used a different address to my own but I couid be tracked by my ip address quite easily. I think talk of Russian spy's etc is nonsense I have nothing to hide and simply use it as a diognosic tool mostly to double check what my cardiologist has said.

A free chat GPT doesn't require any financial input and think how easily we give our bank and credit card details to so many online purchases li wouldn't worry just have some fun with it

ozziebob profile image
ozziebob in reply tofairgo45

Well! It's true, I'm very wary when it comes to trusting others. I don't have either Facebook or WhatsApp because of Zuckerberg and privacy concerns. And would you trust either Trump or Musk? All these tech bros are AI "enthusiasts", if that's the correct description.

As for your advice about having fun, I will do better! Have you seen my new icon photo ... surely dressed to impress and have fun!

fairgo45 profile image
fairgo45 in reply toozziebob

Yes very impressive icon though I had thought you were Australian living in Australia not an Australian living in the UK I bet you turn some heads going to Tescos

ozziebob profile image
ozziebob in reply tofairgo45

I have cornered the cowgirl market over here!😆

I remember those furry chaps from watching cowboy films on TV before you were born.

PS. For the younger members of the Forum, here's a Google AI overview of chaps ...

"Chaps" are protective, over-pants or leggings, typically made of leather, worn over trousers by cowboys and motorcyclists to shield their legs from rain, mud, and potential injuries.

Go figure!

Vonnegut profile image
Vonnegut

Sounds a bit scary to me!

Libby22 profile image
Libby22

I just felt the need to jump in here and point out that AI is being trained on stolen information. Yes, it's clever, but META has accessed a site called LibGen, which appears to be based in Russia, which contains pirated ( stolen) books. META has used these stolen texts to train AI. As an author I'm very unhappy about this and wanted to make people aware of the way these AI applications are being developed in breach of copyright law.

mhoam profile image
mhoam

Jimbobarelli IMHO you need to be very careful about the use of AI, specifically Large Language Models such as ChatGPT to advise on medical issues.

I asked ChatGPT to explore treatment options for a medical condition that I have (not AF) and it provided a very impressive summary of three possible treatments that could be used. Two of the three were just plain wrong and the third was a lot less effective than it stated.

Whilst the mathematics of how LLMs work is well beyond me, they use mathematical modelling to "guess" which words to place one after the other. As stated above, most of this training data has been taken without the author's approval.

It "knows" nothing and "understands" even less. Whilst Machine Learning and AI generally have a lot to offer, LLMs are just a well written bluffer.

Jimbobarelli profile image
Jimbobarelli in reply tomhoam

Seriously I do check it . With my PAF I’ve not found an issue yet - twice it’s pointed to me studies on exercising with AF and my medication which have provided invaluable guidance that I definately wouldn’t have found in Google and probably not my cardiologist either . I find they lump every aF sufferer together - when there is a lot of difference between a 55 year old ex endurance athlete who is still very fit and an 80 year old with underlying heart issues .

Redactrice profile image
Redactrice

Use with extreme caution and fact-check everything Chat GPT says. I saw a few days ago that a Norwegian man is suing the company behind ChatGPT after it falsely said that he murdered his children. This was in mainstream news, not random rubbish on social media. Google it.

fibnum profile image
fibnum

AI info is only as good as what the programmers fed it. I am sure that most of it is well researched, but I have asked AI many time specialized questions which it did not have answers for. Rather than admit that, it gave authoritative, well-written wrong answers!

I would not use answers from AI for sophisticated, complex diagnostics or treatment. It lies convincingly and unashamedly!

This was true of the two free online Chatbots I used.

2learn profile image
2learn

Scary, I believe Ai chat sites are just start of Brave New World/Terminator scenario, people who produce them do not have our welfare at heart only their own profit.

Responsable profile image
Responsable in reply to2learn

Yes, profit + manipulation and brainwashing!

domhnalldubh profile image
domhnalldubh

I wonder what they do with that data you give them?

Jimbobarelli profile image
Jimbobarelli in reply todomhnalldubh

well imagine if it had all the health information on every person and their health issues , with real time data off of their Apple Watch or whatever device .it would surely aid in tackling disease etc ? I know the knee jerk reaction is to think of all the bad outcomes … but it’s a coin with two sides .

domhnalldubh profile image
domhnalldubh

I also wonder if OpenAI anf its major shareholders - Peter Thiel, Elon Musk, Andresesen Horowitz etc - are in the business of tackling disease.

Bennera513 profile image
Bennera513

Absolutely! I am a business intelligence and AI consultant. I also use ChatGPT extensively. Recently I created an Excel sheet of all my blood panel results since 2013. I simply copied and pasted a comma delimited version into GPT and worked back and forth until the data was correct, then I asked what trends or patterns or insights could be gleaned. I then used this info to explore further, in consideration of my symptoms, etc. Then, took this to my Primary and Hematologist.

I also find this very helpful for understanding how different diagnoses are made and how drugs are classified. It's a jumping off point, but the 'conversational' capabilities of AI allow me to have a dialogue that addresses my concerns and allows me to drill deeper. At one point I found AI could also be helpful for emotional counseling (Anxiety/Depression/Fear/ETC).

As others have pointed out you do have to understand that what you are getting is limited to the availability of the data the bots are trained on which will likely only be aged or published studies, many of which remain behind pay wall and therefore, unavailable, generally.

Also, the science and the art of a good response is based on how the question is prompted. I find you can significantly influence the answer and seed the prompt with information that AI will focus on such that you sometimes get the answer you want. Overall a great tool but only as part of a full professional medical review and appropriate common sense.

Having said all that I do also believe that we can slip into a habit loop that has us compulsively and obsessively 'researching' our conditions and monitoring our heart rates,...to the point where we suffer needlessly. I think we sometimes seek certainty and assurance when our physicians have already given us a personalized plan. But as a supplemental source of insight, I do find AI invaluable.

Jimbobarelli profile image
Jimbobarelli in reply toBennera513

All good points . The fact is - as much as the final word is our health specialist - I think most people used to use Google for medical diagnosis/ advice etc . ChatGPT is a gigantic leap forward from that

mjames1 profile image
mjames1 in reply toJimbobarelli

Yes. But the caveat is that Google does not diagnose or figure out anything for you. It simply points you to articles that you can read or ignore based on the credibility of the source and then based on the reading come to your own conclusions. Chat, on the other hand, will often not site sources and will often come to dubious conclusions either because it was trained on outdated or bad data, or that it's just not as smart as it sounds, as of now. I'm certainly not putting it down. It's a great advance and I use it, but you have to be very use the results.

Jim

Not what you're looking for?

You may also like...

AFib and Stroke

Hi guys. Just wondering if you are in AFib for more than. 12 hours and only rely on AAD without...
SpritzerAce profile image

Atorvastatin and Afib

Hi everyone, this is my first Post on here, however, I find all your comments so comforting. I had...
momo24 profile image

AFib and Antibiotics

I know you lovely experienced Afib people will be able to answer my query for me! I have recently...
momo24 profile image

Interesting chat with ChatGPT

I was initially very wary of AI because of some clanging errors I found (a pound of lead was...
Cliff_G profile image

Afib and Escitalopram

Hey all, been really struggling with stress and anxiety since being diagnosed with afib 1 year ago....
Noordy profile image

Moderation team

See all
HollieAdmin profile image
HollieAdminAdministrator
Emily-Admin profile image
Emily-AdminAdministrator
AndreaAdmin profile image
AndreaAdminAdministrator

Content on HealthUnlocked does not replace the relationship between you and doctors or other healthcare professionals nor the advice you receive from them.

Never delay seeking advice or dialling emergency services because of something that you have read on HealthUnlocked.