I’ve been working on this N-of-1 personalized health opportunity/solution for the past five years.
The proposed tools (AI + Health Data = Personalized Nutrition) are already within reach for most people, yet progress remains insignificant.
Why?
For anything to truly change, you MUST do two things: first, disrupt the environment, and second, disrupt the conditions within the environment. Without this two-step disruption, old habits and patterns keep people captive, and neuroplasticity cannot take hold. Same reason why 95% of all diets and New Year's resolutions fail.
This is why you NEED TO MEET PEOPLE WHERE THEY ARE—at home.
Why home?
Because it is the only place where you can effectively leverage the two-step process to instigate REAL and PERMANENT change.
Home as Health: Encapsulating every aspect of human health.
👆I have a proposal for you. Let's find a time to make this a reality.
This is INCREDIBLE. There is so much I want to comment on.
First, thank you for sharing.
Second, I am going to do the same as soon as I am able to in order to share the experience with others. Maybe it will motivate the same exercise, which will bring us one step closer to (metabolic) health.
Third, (perhaps the most important point) it would be interesting to experiment with a prompt that didn't have such rich information (e.g., your success with a ketogenic diet). The reason: I wonder what the baseline recommendation would be - where would OpenAI or another LLM draw advice from? What makes it wise enough to omit studies/guidelines/recommendations deeply rooted in big food and big ag industry ties?
Fourth, the level of *unique* detail is striking. For example, stress reduction techniques can help curb binge episodes. This is what I've been working on over the past year or so. I wouldn't binge junk, I would just binge whatever I would normally eat, perhaps 1.5-2x the amount for no reason, and then feel guilty afterward. When I'm stressed, I'll eat until I'm uncomfortably full - which I don't think is the worst thing given that I eat whole foods and have recently been shifting toward a ketogenic diet - but still, it's not a GREAT habit/tendency to have.
Which brings me to five, the ketogenic diet. I feel incredible ever since first, eliminating UPFs from my diet years ago, and again, (it's like health^2) ever since I started eating less carbohydrates and more fat, about 2-3 months ago after looking into some of the research (which I have yet to do a proper dive into). My energy and mood are better and more stable, I feel lighter physically and don't need to eat so much or so often (I think this is due to the fat.)
I still enjoy rice, sourdough bread, carbohydrate-rich vegetables, etc. It would be interesting to experiment with a strict keto or even carnivore diet, but I would need to find the motivation and get "unstuck" from my bias against restriction. The same goes for intermittent fasting. I love breakfast, but noticed on days with back-to-back early meetings that if I only have coffee, I'm not hungry. But I put a good amount of cream in my coffee, so who am I kidding?
I am open to trying some more prompts. I was trying to integrate another genetic summary that gave me diet recommendations but could not figure out how to get it into the system and the genetic datasets are wobbly.
I could see having a persistent dialog in a gpt format. Maybe I should take picture of food to see if it logs it.
Yes! Like, it could become a nutritionist supporting diet-related chronic disease *reversal* -- the dietitian that I assume is a major cost of a lot of nutrition guidance startups (e.g., Virta). Give patients the option to upload photos of food for ease.
It's interesting that this seems to have done a decent job, at least based on a quick read. My prior would be that ChatGPT is basically sourced from a lot of dubious web material funded by big pharma/food/ag, and therefore that the result would be regurgitation of previous bad advice.
Maybe the symptom of that is that it barely mentions fiber or green vegetables.
As a physician, this article served as a great reminder of the individuality of every patient. With all the population-level data, it's easy to lose sight of the fact that each person’s needs, physiology, and response to treatment are unique. N-of-1 care underscores the importance of tailoring treatment to the individual while reinforcing the core value of personalized, patient-centered care.
Thanks for sharing! I have completed the bloodwork for several years via Inside Tracker and have worked on my own to develop my response. Duh, this is fantastic.
When our dog was sick during the summer and had a bunch of blood tests and other tests I was able to upload all that into chatGPT and get a much better summary of the meaning than I was getting from the vet.
I asked it to explain to me the situation and options, rather than ask for a diagnosis. When you ask it to make a specific diagnosis, ChatGPT falls short. But if you ask it to explain the situations and what other symptoms to look for, it was very helpful and comforting.
I find the same is true with the human blood tests
Carter, we should talk…
I’ve been working on this N-of-1 personalized health opportunity/solution for the past five years.
The proposed tools (AI + Health Data = Personalized Nutrition) are already within reach for most people, yet progress remains insignificant.
Why?
For anything to truly change, you MUST do two things: first, disrupt the environment, and second, disrupt the conditions within the environment. Without this two-step disruption, old habits and patterns keep people captive, and neuroplasticity cannot take hold. Same reason why 95% of all diets and New Year's resolutions fail.
This is why you NEED TO MEET PEOPLE WHERE THEY ARE—at home.
Why home?
Because it is the only place where you can effectively leverage the two-step process to instigate REAL and PERMANENT change.
Home as Health: Encapsulating every aspect of human health.
👆I have a proposal for you. Let's find a time to make this a reality.
https://www.linkedin.com/pulse/home-health-first-frontier-healthiest-humans-3xftc?utm_source=share&utm_medium=member_ios&utm_campaign=share_via
This is INCREDIBLE. There is so much I want to comment on.
First, thank you for sharing.
Second, I am going to do the same as soon as I am able to in order to share the experience with others. Maybe it will motivate the same exercise, which will bring us one step closer to (metabolic) health.
Third, (perhaps the most important point) it would be interesting to experiment with a prompt that didn't have such rich information (e.g., your success with a ketogenic diet). The reason: I wonder what the baseline recommendation would be - where would OpenAI or another LLM draw advice from? What makes it wise enough to omit studies/guidelines/recommendations deeply rooted in big food and big ag industry ties?
Fourth, the level of *unique* detail is striking. For example, stress reduction techniques can help curb binge episodes. This is what I've been working on over the past year or so. I wouldn't binge junk, I would just binge whatever I would normally eat, perhaps 1.5-2x the amount for no reason, and then feel guilty afterward. When I'm stressed, I'll eat until I'm uncomfortably full - which I don't think is the worst thing given that I eat whole foods and have recently been shifting toward a ketogenic diet - but still, it's not a GREAT habit/tendency to have.
Which brings me to five, the ketogenic diet. I feel incredible ever since first, eliminating UPFs from my diet years ago, and again, (it's like health^2) ever since I started eating less carbohydrates and more fat, about 2-3 months ago after looking into some of the research (which I have yet to do a proper dive into). My energy and mood are better and more stable, I feel lighter physically and don't need to eat so much or so often (I think this is due to the fat.)
I still enjoy rice, sourdough bread, carbohydrate-rich vegetables, etc. It would be interesting to experiment with a strict keto or even carnivore diet, but I would need to find the motivation and get "unstuck" from my bias against restriction. The same goes for intermittent fasting. I love breakfast, but noticed on days with back-to-back early meetings that if I only have coffee, I'm not hungry. But I put a good amount of cream in my coffee, so who am I kidding?
Thanks again for sharing.
I am open to trying some more prompts. I was trying to integrate another genetic summary that gave me diet recommendations but could not figure out how to get it into the system and the genetic datasets are wobbly.
I could see having a persistent dialog in a gpt format. Maybe I should take picture of food to see if it logs it.
Yes! Like, it could become a nutritionist supporting diet-related chronic disease *reversal* -- the dietitian that I assume is a major cost of a lot of nutrition guidance startups (e.g., Virta). Give patients the option to upload photos of food for ease.
Interesting discussion early in the podcast: https://www.youtube.com/@allin
It's interesting that this seems to have done a decent job, at least based on a quick read. My prior would be that ChatGPT is basically sourced from a lot of dubious web material funded by big pharma/food/ag, and therefore that the result would be regurgitation of previous bad advice.
Maybe the symptom of that is that it barely mentions fiber or green vegetables.
As a physician, this article served as a great reminder of the individuality of every patient. With all the population-level data, it's easy to lose sight of the fact that each person’s needs, physiology, and response to treatment are unique. N-of-1 care underscores the importance of tailoring treatment to the individual while reinforcing the core value of personalized, patient-centered care.
Thanks for sharing! I have completed the bloodwork for several years via Inside Tracker and have worked on my own to develop my response. Duh, this is fantastic.
When our dog was sick during the summer and had a bunch of blood tests and other tests I was able to upload all that into chatGPT and get a much better summary of the meaning than I was getting from the vet.
I asked it to explain to me the situation and options, rather than ask for a diagnosis. When you ask it to make a specific diagnosis, ChatGPT falls short. But if you ask it to explain the situations and what other symptoms to look for, it was very helpful and comforting.
I find the same is true with the human blood tests
having fun with O1 Pro, I see.
really interesting article - how much did the testing cost?
$750 It was a lot of testing