1. < Epilepsy & Seizures
Profile picture for Dan Noyes

Building Cognitive Scaffolding with AI

Posted by Dan Noyes @dannoyes, Aug 17 11:04am

I'd love to know whether anyone in the group uses AI to manage their epilepsy, including to create a framework for any cognitive challenges they may face. I'm not talking about using AI to manage the clinical side of your epilepsy or seizure disorder, but to use it as an aid to daily living.

For example, I use a seizure-informed AI framework with Claude and ChatGPT to provide me with assistance in my daily life planning, supporting my memory, which has been compromised due to my multiple TBIs. This has been a huge help as I manage my symptoms.

While it will never remove the challenges we face, I've found it helps me overcome some of the day-to-day frustrations of living with epilepsy. In one case, it even corrected me when I was pushing too hard on a project. "Dan, you've been working on this task for three hours now; you need to take a break," Claude's gentle warning said.

Here are a few articles that might spark a few ideas: https://www.cureepilepsy.org/seizing-life/epilepsy-identity-and-ai-processing-the-impacts-of-an-adult-onset-diagnosis/

Interested in more discussions like this? Go to the Epilepsy & Seizures Support Group.

Profile picture for Chris Gautier, Volunteer Mentor
Mentor
Chris Gautier, Volunteer Mentor | @santosha | Aug 26 8:44am
In reply to @dannoyes "@santosha isn't it crazy to realize how much we've all learned about AI in one year...." + (show)
Profile picture for Dan Noyes @dannoyes

@santosha isn't it crazy to realize how much we've all learned about AI in one year. Here are some thoughts about AI with out medical journey which I think is really important for anyone with a chronic medical condition.

Seven Principles of Safe Healthcare AI (the first two are things not to do). You will also find that some points have an experiment you can try on your own.

1.) AI should never replace direct human engagement. While we all know this, I feel it is important to create guardrails and boundaries about how much time we spend with AI and to understand the risks of building "relationships" with our technology. As AI agents build a deeper contextual understanding of us, they will continue to feel more human-like, so we need to establish safeguards for our emotional health. Having direct human connection can never replace the hug from someone we love. [Go to ChatGPT and ask: "What are three things I can do nurture my relationships with my family and friends?"]
2.) Never share information that you wouldn't want to share with your next-door neighbor. We must always remember AI chats are not private. So, practically never share your address, your birthday, phone number, family information that is not public, or I would go so far as to be careful about medications you take that you wouldn't want someone to know. Be careful about oversharing. [Go to your favorite AI tool and ask "What do you know about me?" Be ready to be surprised.]
3.) Use AI to help you prepare for appointments. I have my AI tools provide me with a one-page summary of my current issues, trends, and observations of my epilepsy. I then take that sheet to my appointment and hand it to my provider. I have found they love the summary. [Ask your AI companion: "I have a six month follow-up with my provider. Here are my concerns. Help me compile a single-page summary I can give to my doctor for our visit. If there is something you think I should include that I haven't told you let me know"]
4.) Have AI identify trends in your condition. AI is great at connecting the dots. So I try to provide data it can use to help make sense of my condition. For example, through my Apple Watch, I have my sleep data, which, when combined with my exercise, can help to give some observations I can share with my provider.
5.) Lab work analysis. I take my bloodwork results and let AI provide me with an amazing analysis and explanation of the trends it sees. However, the real value comes when I share that perspective with my specialist, who can give a clinically accurate assessment of the results. [I prefer Gemini for this. "Here is the bloodwork from my last lab test. Help me understand the significance of the results. What, if anything, can I do to improve the results? Here are also my previous results. Do you see any trends that I should review with my doctor?" Remember, in most cases you have 12-15 minutes with your provider so make it count.]
6.) We are all advocates for our healthcare so I find AI is an amazing tool for researching the latest scientific work being done in each of our presenting symptoms. AI becomes my research partner. [You can use any AI tool for this, but I tend to like ChatGPT or Claude for this: "Act as my personal medical researcher. What are three current trends for those who suffer from (your type of seizure disorder). Only include peer-reviewed journals in your research. Then write a report for me that is written for someone in the sixth-grade to help me better understand my condition"]
7.) Embrace the role AI provides you as a sovereign of your own healthcare. For years, I felt alone and bound by my condition and poor healthcare. Six months ago, one provider suggested I go on a medication that would have had life-altering side effects. Really bad stuff. AI gave me a sense of agency over the possible risks this medication presented. When I shared these concerns, my provider was upset and tried to "fire me" as a patient because I questioned his recommendation. Fortunately, my AI research combined with the assurances of my Mayo Clinic team provided me with the confidence to assert my rights as a patient and quite frankly as a human being.

There are many critics to AI in our care, but it is another very strong tool in our arsenal as we manage our chronic medical conditions.

Jump to this post

@dannoyes
Hi Dan,
Thank you so much for these genuinely great tips!
On the privacy point, I actually share quite a bit more with my paid version of Claude than I would with my own neighbor, since I understood that paid AI versions keep everything I share private.
Separately, I recently started using Projects within Claude for each topic I ask for help with, and it's made a real difference in keeping things organized.
Am I wrong about the privacy piece? And do you use Projects too on Claude? If so, I'd love to hear how you structure yours, or any tips you might have.
Thank you again!
Chris

REPLY
Profile picture for Dan Noyes
Dan Noyes | @dannoyes | Aug 26 9:50am

@santosha I'm so glad you asked about Claude. The only accounts that are truly private are commercial accounts designated as Claude for Work or Enterprise/Teams. You can opt-out of having your data shared by manually going to Account Settings > Privacy and turning off the option to "Help improve Claude..." This raises a much bigger privacy issue that you have to answer personally and intentionally, considering the ramifications. I find that I answer this question by asking myself, "What information have I already shared in a public venue, hence it has already been made public?" So, for example, since I have written a book recounting my epilepsy journey and have shared mentions of my "neurological condition" on LinkedIn, it is already available in public. I made this decision intentionally, but I did so understanding the risks and exposures. What is true for me is not true for others. The biggest point is knowing that nearly all AI tools are not private. However, my doctor's office has very thin walls, so I can hear everyone around me talk about their medical condition. Those are not private conversations, even though people assume they are.

I try to think of my epilepsy and brain fog as a data storage issue. My brain isn't processing data correctly and, in some cases, not at all. The consequences of this data storage issue is that I forget things, some of which are important. I have forgotten family vacations. I can't remember the names of people around me; I often don't remember when I've said something in the past, so I know I may repeat myself. In the end, my brain suffers from a data issue. With that understanding, I have to find a data management solution. Once I realized this, I could start finding solutions.

So like you said Claude projects are great as are tools like Gemini Gems. I may also use ChatGPT's "GPTs". Both Claude and ChatGPT have contextual memory. They understand the big picture and context of my life. This is a huge bonus and once you see it in action it is hard to go back. Once again just query any of the tools with the prompt "Tell me everything you know about me" and you might be amazed.

Since I'm not always at my computer I use tools like Bear and Drafts as capture tools on my phone when I think of something important I don't want to forget. I also have used tools like "Flow" and "Audio Pen" to supplement my notes.

If you want a more expanded way to capture a private, big picture of your life, your healthcare journey you may want to consider a tool like Obsidian which helps you capture really everything of value that enters your brain and then see the way it all comes together. This is not for everyone, but it can be a huge help if you want to build a neural data picture of your thinking.

The seizures of epilepsy suck. They are not for the faint of heart. I find the biggest challenge is not the seizure activity, but the way it impacts my memory and neural processing is where I find the biggest challenge. I take the neuro-psych tests at Mayo and they can see what I feel. There are tricks to help improve your recall, but those only go so far. I met a new neighbor yesterday. His name is Lou. So my mind associated him with the character who played the original Hulk (Lou Ferrigno). Now when I see him do I immediately see the Hulk? I sure hope I don't say "You sure don't look as green as I remember you."

REPLY
Please sign in or register to post a reply.