As you might have guessed, I do a lot of work with AI. Having it help me with my podcast, writing my show notes—it never writes my podcast. It just helps me to get text out. I’m not a particularly good writer.
But then I started using it more and more, trying to figure out how I can use it more in my personal life. And what I’m going to talk about is how I’m using it to help my health.
Keep in mind, I would never say that AI should replace your doctors or make important decisions for you about your health, about the medications you’re taking. We’ve all seen horror stories where people have used AI and trusted it too much. It can definitely make mistakes, and it is just one kind of point of view. This is not about blind trust at all.
There are some ways, though, that this past year I have used AI successfully in helping me make health decisions, while still staying firmly in control over my own choices and giving me great launch points so that I can talk to my doctor.
## Losing 80 Pounds and Coming Off GLP-1
I will give you the background. I lost 80 pounds in the last 18 months, and now I’m trying to come off the GLP-1 drug I had been taking, which helped me lose weight. I’ve been struggling with weight my entire life, and if you’ve looked at any of the studies, many of the people who stop taking GLP-1s gain weight back. So I’m thinking, can I use ChatGPT to help me in keeping this weight off?
## Using ChatGPT for Daily Food Tracking (Without a Diet App)
Some important factors to know are that when you have lost this much weight, you need a ton of protein to keep your muscles strong because muscle loss is a part of any big diet. You want to make sure you have enough fiber, you have enough water, and you get enough exercise. But too, I need a plan about how I am going to come off this drug in a very careful way because I don’t want to put the weight back on, and I’m very, very frightened of that.
One of the things is that when you’re on any sort of a diet plan, if you’re tracking it or you’re using a dieting app, it’s not really built for you, and it has some very rigid rules in it that usually don’t tend to be good for long-term maintenance. My situation is very unique. I just dropped 80 pounds. My body is trying to put back the weight because it thinks something terrible just happened to me.
And what AI is able to do is start recognizing patterns in my own life. I can use it to say, am I getting enough food for my maintenance plan?
## Meal Feedback and Macro Adjustments
Essentially, I looked at how many calories I should get in a day from ChatGPT, and I decided to increase my calories 50 calories a week so that my body can get adjusted to it. And then I started telling it what I eat—what my breakfast is, what my snacks are, what my lunch and my dinner are like—and it keeps track of everything for me.
## Meal Feedback and Macro Adjustments
It’s able to tell me how much fiber I had in a day. Did I get enough calories today? Is the fat that I’m getting in my meals supporting me, or maybe is it dominating my calories too much?
I thought about it, and what a lot of people do—and the people who have the most success—is track their food using a food nutrition program. They’re great. I think they’re fantastic. I don’t like micromanaging my life that much. I found it very draining to track every ounce of food that went into me.
Now it’s fun and it’s great, and I gave a review for the app I think is the best, which is My Net Diary. But I was hoping that there’s a way I can maintain this weight loss without tracking my food.
So every day I tell ChatGPT what it is I ate. It will give me the stats—how much fiber, how much protein, how many calories—in the ballpark. But it also offers advice.
So it might look at one of my meals, namely breakfast, and say, I saw that you mentioned that you had toast and two eggs. Did you eat the toast before the eggs or after the eggs? And I thought, that’s an interesting question. And I said, well, before the eggs, because it’s quicker to toast than it is to cook eggs.
And it suggested that I would digest things better if I ate the protein first and then the carbs second.
Or maybe I just had a disastrous day. I ate something I shouldn’t have eaten. I went out with my friends and I ate too much. It can give me that breakdown of the day. I can tell it, you know, I have one more snack left in the day. Is there any way that I can make this day better, even though I kind of screwed it up today?
And it would say, you know, having a cup of cottage cheese would be a good move. It would give you some of your missing protein that you didn’t get today with very few calories, and it would bring your day into more of a maintenance kind of style.
It can also see trends. I started eating nuts, and it started getting concerned that the nuts were overwhelming the amount of fats I was having in a day. And it said, this would be fine except you had fats here, here, and here. So maybe either cut those fats out and have the nuts, or reduce the amount of nuts you’re eating.
So it gives me very valuable ideas. No all-or-nothing kind of thinking involved in it. It just tells me how I can adjust.
## 30-Day Trend Analysis for Nutrition
It can look back, and I can ask it, hey, over the last 30 days, how’s my eating going? Am I on target for what my goals are? And it’ll say, well, your calories have been good. You’ve been getting the right amount of calories, but maybe you have to increase your protein a little bit.
Now actually, my protein is pretty good too. I’ve been having trouble getting fiber in. So then it would say, you know, pears have a lot of fiber in them—four grams. Not a lot, a lot, but enough. Or you could try eating this or this or this. And it will give me a number of suggestions about meals I can swap out from my last 30 days of eating so that I can make my next 30 days a little bit better.
## Building a Flexible Exercise Plan
But even when it comes down to exercise, it can help me build an exercise plan. I go to a trainer on Monday, Wednesday, Friday, but on Tuesday, Thursday, Saturday, and Sunday, I’m on my own. So it helped me develop an exercise plan that would be pretty good for me.
## Exercise Trend Tracking and Weekend Struggles
I told it I’m struggling. I don’t feel like exercising. And it gave me some gentle ways that I could get back into the exercise world with cardio in slow steps and try to achieve that a little bit better.
But you can also go to ChatGPT and tell it, you know what, I did not sleep well last night. I don’t have a lot of energy. Is there something that I could do that would be beneficial to me now and not so aggressive as having a full-blown cardio workout?
And it would find a way. Maybe do a slow row. I have a rowing machine. You know, get a slow row in for a while and do some stretches.
Or at one point a couple of years ago, I hurt my ankles and I was pretty laid up. I couldn’t walk without a limp. I couldn’t figure out how to exercise without causing my ankles to hurt more. And so I tried it out a few weeks ago. I told it what injury I had to my ankle tendons. What could I do for cardio if I’m in this situation?
And it gave me a bunch of ideas that I could do without actually damaging my ankles. So AI would have helped me a great deal had I had it when I screwed up my ankles.
It is about coming up with an exercise plan, I think, that isn’t perfect. Not every day you feel great. Not every day your ankle feels great, or maybe your back hurts or something like that.
Also, because the app has notification ability, I asked ChatGPT to every day ask me if I exercised. Allison’s doing a good job of this too. But then it records what I tell it. And so I can say, over my past 30 days, can you notice any trends about my exercise?
And it said, you really are doing a pretty good job on Tuesdays and Thursdays, but on the weekends you kind of fall apart. Let’s come up with a plan on your weekends that we can make a little bit easier for you. What do you think the problem is?
And I said, well, the problem is that I get up to go to work and there’s a structured time right there to exercise before work. But on the weekends, I sleep in a little bit more. I don’t have a structured day. And then I just kind of forget about exercising.
So I came up with some ideas to make me do a little bit better.
Let me give you one last use case when it comes to health and how I’ve been using it.
So with the process of losing 80 pounds, my lab results are really all over the place. I used to have high cholesterol, high blood glucose. I used to have problems with my blood pressure.
And I was able to feed my stat numbers in from my doctor’s appointments and get the idea—am I headed towards normal? Am I already at normal? Or is there something that I should be doing, drinking more water, something that I should do to make my numbers better?
Now in my case, my numbers got entirely better. I am actually a perfectly healthy human being now.
One of the interesting things is I used to have high blood sugar in the morning. Now I have low blood sugar in the morning after dropping 80 pounds. So it has been helping me come up with a plan to get my glucose up in the morning so that I don’t start going on a roller coaster ride with my blood sugar. And it’s been helping out. And all it took was eating food in a little bit different order.
So it can actually tell you, when you give it lab numbers, what good ranges are, what the test is actually measuring, whether the fluctuations matter at all or if it’s just kind of noise, and where there are tests that have multiple related values.
So for example, one of my kidney numbers came up too high. But then ChatGPT said, but this second number—if that number was high too, I would say you were having a problem with your kidneys. But because that stat is low, it’s a good indication you were just a little dehydrated. Talk to your doctor about it.
It was never shying me away from that. But instead, it gave me a little peace of mind that probably my kidneys were okay. And then when the doctor came in later, you know, a couple days down the road with his analysis of my blood labs, he said exactly the same thing.
## The Eye Scare: Optic Nerve Swelling
So right before Christmas, I went to my eye doctor. And this is just a regular eye doctor that you might get at a corner store kind of place. And she said, I can see that your optic nerve is swollen. A lot of things can cause that, but we need to refer you to a retina specialist.
Boy, you want to talk about a scary situation.
The first thing that ChatGPT was able to help me do is I asked it what could cause this. I’m on a GLP-1. Is the GLP-1 causing it? Because there is some concern that GLP-1s caused this problem in other people. Turns out it’s not true.
But it said chances are you were diabetic for a while, for about five years, and it’s probably damage that you did while you were diabetic, which is really scary because I wasn’t that diabetic. I just basically had high blood sugar in the mornings.
It kind of put it more into reality. It was most likely from diabetes, and it was most likely something that would get better. So my panic kind of came down.
Obviously, it was never going to tell me not to go to the doctor. I did go to the retina specialist. But essentially it gave me some peace of mind.
## Preparing for the Retina Specialist
The second thing it did for me is it told me what I could expect at that retina specialist eye appointment. It listed accurately the tests that I was going to take, how much time each test took, and even made a nice suggestion that said you should bring your AirPods along. Sometimes there’s gaps in between these tests and maybe you might get nervous, and having your AirPods along would help you stay a little bit more calm.
That was a great suggestion. And thinking, what kind of search engine tells you to bring your headphones along so that you don’t get nervous? That was kind of amazing.
## Asking Better Questions at the Doctor
Then the last thing it did is it gave me this idea of what kinds of questions I should ask the retina specialist when all my tests are done. ChatGPT even indicated that the retina specialist will most likely say this phrase, that this is what this is. And if he says that phrase, you want to ask this clarifying question: is the macula damaged?
And when I got to that point, sure enough, the doctor said those words. And I asked the question, does this include macular damage? And he said, well, that’s a really incredible question.
And I said, to be honest with you, ChatGPT told me to ask you that.
But ChatGPT gave me five really great questions. I memorized them, but I also had them on my phone if I needed to. And the doctor was very impressed with the questions I asked.
The end of it is that it was caused by diabetes and that it’s going to come out and fully resolve itself. So it was a huge relief to me.
But using ChatGPT completely helped me from getting out of control with stress about this issue, gave me practical advice for the meeting, and then told me good questions to ask my doctor.
## Apple + OpenAI + HealthKit Integration
Coming up, you’ve probably heard the news that Apple signed a deal with OpenAI so that there will be an integration between HealthKit, which stores all our information—how much we’re walking, whether we have asymmetry, what our VO2 max is, depending on if you have a watch—a lot of details about our health on there. And ChatGPT will do a deep dive analysis on some of that data.
## Health Data Permissions and Privacy Concerns
I’m on the waiting list for that, even though a lot of people out in the tech podcasting world felt that that was maybe ill-advised.
But rest assured that, first of all, Apple is very secure. They will not allow you to do something without your express permission. These aren’t going to be automated recommendations. It’s not going to be continuous monitoring. And it’s not going to have full access probably to your medical records unless you specifically give it that. I don’t even know if it’s going to ask for that.
The information is going to be read-only. It’s going to be opt-in. And it’s going to be only very specific categories. And ChatGPT is going to be focused on trends. It’s not really looking to tell you what your blood lab looks like. It’s going to talk more about your glucose is coming down. Your exercise minutes are coming up. Normally, if you don’t get in a thousand steps before 10 o’clock in the morning, you are going to be very low for the rest of the day. Do you think you could maybe go for a walk? Maybe looking at some of your activity summaries.
While you’re looking at this particular data, you will not, I believe, endanger most of your data. I think Apple is going to be very careful about it. It’s going to do permissions exactly how it does permissions today.
When we sign up for a new diet app or a new food tracking app or a new exercise app, those apps ask you for various permissions to see certain kinds of data or store certain kinds of data. This is not going to store any kind of data.
But you do give permission to third parties that you don’t know what they’re doing with the data necessarily.
I remember once I downloaded an exercise app from the Apple Store and it asked, you know, some questions like, can we see how much you’re walking? Okay, sure, you’re an exercise app, fine. Can we see your heart rate? Okay, that makes sense too.
And then it asked me for information that was none of its business. Why in the world would an exercise app want to know this? And I said no.
So we know already in HealthKit we have very granular control over our health data, either allowing or denying other apps from seeing certain kinds of material.
And we also know that if OpenAI did anything that wasn’t based on the permission, wasn’t given complete consent to, Apple would come down on that company, remove their access entirely. They are very security conscious.
So I have no doubt that this is going to be very well managed.
There also is some indication from the early people who are part of the testing that this is an isolated instance of ChatGPT. So when it looks at the data, it’s not going to remember what it said in the past. It’s not looking at conversations you had with it in the past. It’s not going to remember I told it I had an eye problem. It is a secured, separate conversation that is just going to look at the data that it’s supposed to look at.
And I get it. I understand why skeptics are worried about it. Security people are worried about it. And, you know, I don’t want to share my health data with an AI, and that’s a completely valid decision that you might have.
But I think it’s a misunderstanding of how security with HealthKit works already. It’s not all or nothing. It’s something you can reverse at any moment. It’s very open and visible about what you’re giving permission for it to do.
And it’s not going to be used for training, you know, when they talk about AI training.
## Future Potential of AI Health Insights
We have to see what it’s going to be. We don’t know quite yet because it’s not out for consumers yet. But I think it’s going to be something that can be very helpful to give us insights about how we eat and how we sleep and how we exercise and whether we’re going up or we’re going down or whether my ankle stability is getting better or getting worse.
I think it’s going to be very valuable.
A few years ago, I did a review about my sleep app that I use to help me fix my sleeping because we got into a debate with Allison. Do sleep tracking apps matter? Are they valuable to us?
And I had to pull all that data out, put it into Excel spreadsheets, and do my own correlation testing. I found some valuable things out.
## Sleep Tracking and Manual Data Analysis
I found out I wake up every day at 3 o’clock. It turned out it was my water softener making noise at exactly 3 o’clock every night.
I also have trouble sleeping on Tuesday nights and Saturday nights. Why? Because I go to my friend’s house. I’m an extrovert. I get all wound up. And as soon as I started realizing that going over to my friend’s house was causing me to sleep not great, I started doing some cool down, like getting my mind to cool down a little bit so that I could actually fall asleep easier on the nights where I got too much social interaction going on.
Wouldn’t it be great if an AI could help me analyze those trends without me having to dump everything out into Excel and do my own statistical analysis?
So I think it’s going to be great.
## Why AI Pattern Recognition Matters
And right now, like I said, for health issues, AI has helped me analyze patterns with my food, my exercise, my sleep, my health overall.
There’s no emotion there. There’s no stopping point. Right now when we go to a doctor, we get a limited number of questions we get to ask right off the bat. And then half the time, if you’re anything like me, you go home and you think, wait a minute, I have like five more questions.
AI is an endless supply of question answering. It doesn’t bill us for more questions. It doesn’t get frustrated with us. And it helped me, in my situations, prepare to talk to my doctor in a more intelligent way.
Having that kind of resource is valuable.
## Comparing Different AI Tools
The one kind of thing that I did in part of this is that ChatGPT tends to be very friendly. If you take a look at all the various AI apps out there, it’s a chat AI. It’s very friendly, very encouraging, and very positive about things.
And if you take a look at Grok, it is a very good devil’s advocate, arguing whatever point you’re trying to make to make sure your thoughts are in the right place.
If you’re looking at Claude, it’s a fantastic writer and a really good coder.
If you look at Copilot, it’s about business and work and those kinds of things.
And if you look at Perplexity, Perplexity is very science-based. It is cold. It’s not friendly at all. It’s not unfriendly either. But if you ask it a very specific question, it will give you scientific research or backing to your questions.
So if I went to it and ChatGPT said your optic eye swelling is probably this, I could actually put those words into another AI—I use Perplexity—so that it can say this is what those words mean, and here are my references so that you can read more about it on your own. Very scientific.
Also, I mentioned Gemini. Gemini has the personality of Google as an AI search engine. So it is also not as friendly, not as chatty, but also can give very good advice because it is a search engine in the end.
## Cross-Checking AI With Other AI
But having one AI and then being able to bounce your ideas off of another AI is another good way that you might be able to check out the information the AI is telling you.
So if you’re told your kidney score is this, you might go to Perplexity and ask that question, and it will give you articles you could read as a reference and a summary of those articles to give you a little bit more information.
## Final Thoughts: Informed, Not Replaced
We’re not getting a diagnosis, and we’re not going to change our medications because AI told us to. Instead, we’re going to be getting clarity about what we’re already being told, getting definitions for words we maybe don’t even know what they mean, and like I said, walking into our doctor’s appointments more informed and more educated about what it is we are dealing with and good questions we should ask that doctor.
For me, it’s been incredibly helpful. And hopefully it’s going to help me figure out too how to keep this 80 pounds of weight loss off.
[…] AI is Changing My Health […]