The World’s Best-Prepared Patient

I have a good friend who lives in Denver. He was sick for months: coughing, shortness of breath, a fever that wouldn’t quit. He saw half a dozen doctors, and none of them could figure out what he had. Then he went to National Jewish Health, here in Denver, and right away they diagnosed him with Valley fever, a fungal lung infection caused by inhaling a fungus that lives in dry, dusty desert soil.

Valley fever is not prevalent in Denver. My friend spent winters in Arizona and caught it working in his garden (note to my wife: another reason why I refuse to garden or work in the backyard), and the Denver doctors had never seen it. National Jewish Health is known as a national (key word) authority on pulmonology and sees patients from all over the world, so one of the first questions its doctors asked was where he had spent the past year.

He told me this story ten years ago, and it stuck with me. Half a dozen well-meaning doctors could not help my friend because their knowledge was constrained by their myopic (geographic) circles.

That was before AI.

More recently, another friend had cataract surgery. A few months later he started losing his vision. He saw one doctor, then several more, and they all recommended eye drops.

Then he described his condition to AI in great detail. The AI offered a different diagnosis and told him he needed to see a doctor immediately and might need emergency surgery. He took that diagnosis to a new doctor, who examined his eye through the lens of the AI’s suggestion and agreed. He had emergency surgery, and his eyesight was saved. When I asked him why the other doctors had missed it, he thought it was a combination of things: poor diagnostic equipment, incompetence in one case, and an uncommon infection masking the underlying problem.

These two stories made me pause and question how I approach my own health.

I have a lot of respect for doctors. Being a doctor is truly one of the noblest professions, and becoming one requires a huge sacrifice. They do more good for society in a week than I do in a good decade.

But they are… well, they are human.

An average doctor sees a few thousand patients a year. No doctor could remember all their names, let alone the full extent of each patient’s medical history. Doctors are overworked, and so the doctor-patient visit resembles a pit stop in a Formula One race. In the few minutes you get, your doctor must reconstruct you from notes written months (and five hundred patients) ago, or even worse, from the generic questionnaire you filled out in the waiting room while squirming in an uncomfortable chair.

Although they are calm and professional, wear white coats, and may appear to be godlike figures, they are not immune to whatever happened to them that morning: a spouse who woke up on the wrong side of the bed, a son whose heart was broken by his longtime girlfriend, or even a rude patient an hour earlier. Life walks into the exam room with them.

It struck me that my health is ultimately my own responsibility. Nobody cares about our health more than we do, except maybe our mothers. We know our bodies and our medical history better than any doctor ever will. But we lack the knowledge a doctor has, and our memory is not perfect either.

As I thought about this, I realized that nobody in this equation is flawless: not doctors, not us, and not even AI. It hallucinates, making up stories like a Hollywood screenwriter. But it also has perfect recall and lacks the myopic vision of human beings. It has access to an up-to-date medical library and the latest discoveries, infinite patience, and no mortgage payments. It has no bad morning riding into the room with it.

By now, you are probably thinking I am battling some sickness. Thankfully, just the opposite is true. I am in the best health of my adult life. I am closer to my ideal weight than I have been in years; I lift weights twice a week and do yoga twice a week, and I walk in the park nearly every day. I eat only whole foods, and I consume little alcohol or sugar. But I walk under the same sun as everyone else, and I have no idea what cards the health gods have in store for me.

Our health depends on seven factors: diet, exercise, sleep, stress, relationships, genetics, and luck. Five of the seven I can control. That matters to me, because I want a long life, but more than that, I want, if I can pull it off, a long, healthy life free of sickness and pain. Our healthcare system is focused on treating sickness; my focus is on staying healthy and catching problems early, when they are easy to treat.

So here is what I did.

I created a project in Claude. A project is a silo where everything I share is saved and retained in the AI’s memory across conversations. (ChatGPT has a special Health project, which I am not using yet.) I use the paid business version of Claude, so my information is not used to train its models.

I connected Claude to my email, my calendar, my digital scale, and my Oura ring, which tracks my sleep, resting heart rate, how much I walk daily, and a dozen other metrics around the clock.

I did not stop there. I gave Claude access to my doctor’s portal, and it downloaded my medical records from the past eight years: blood tests, medications, supplements, doctors’ notes. Everything.

Why? Four reasons.

First, it monitors my health. Every Formula One car is hooked up to hundreds of sensors and analyzed by a team of mechanics, and I’m sure by AI as well. I am doing the same thing, except the analysis no longer happens only once a year during a pit stop at the doctor’s office. Once a week, the AI runs a routine that analyzes my vitals from all sources and gives me a health report card. Every few months I get a blood test done, and the results go into the same picture.

Once a week is probably the right cadence for me. Looking at your health every day is like looking at your stocks every day: you’re mostly exposed to noise, and it would make me obsessed with my health.

Other than a one-time setup and occasional maintenance, feeding the AI data takes little time or energy. But compounding doesn’t exist only in finance; it exists in health too, and it happens to bad things as well as good ones. The weekly review helps me identify areas to focus on. If I am not getting enough sleep, my resting heart rate is creeping up, or any of dozens of other markers from Oura and bloodwork start trending the wrong way, I can make corrections. Sleep has been my weakest link, and now I have something that keeps reminding me.

Second, I now have a health databank. Data from different parts of my life lives in one analyzable, objective place. If my health steps on an unlucky rake, I might detect it sooner. For example, Oura showed that my resting heart rate rose while I slept. Claude looked at my calendar and said it may have been caused by higher elevation: I had driven to Vail, which sits nearly 3,000 feet higher than Denver. Nothing was wrong.

Third, the AI sees trends and has greater context. It looks at eight years of a number and asks whether it is drifting, how fast, and what other numbers have moved with it. It can answer many of the questions a doctor asks with more precision, because it is answering from the full record on my computer rather than from memory and a few minutes of chart-reading in the exam room.

Fourth, I no longer walk into a new doctor’s office as a stranger. I bring a résumé of my body: every test with its trend, what other doctors have diagnosed, and what changed since my last visit. Doctors, like the rest of us, anchor on the first thing they read.

That is why I intentionally put the numbers first, the diagnoses by other doctors next, then what changed since last time, and my questions and the AI’s questions last. I want my doctor to form her own opinion, not to be influenced and steered in the wrong direction by this document. My doctor’s pit stop hasn’t gotten any longer. But those minutes are now spent making decisions instead of catching up.

It’s true that AIs hallucinate. That’s harmless when you ask for the capital of Morocco (not Marrakesh, but Rabat, just FYI), but not okay when it comes to your health. That is exactly why I don’t rely on just one AI. There is another very important nuance here: when it comes to my health, I always use the latest, most advanced AI models at the longest thinking depth. This ensures that instead of a first-year medical student doing the research, I have a PhD doing the work. The quality of the model matters in the answer you get.

We are used to asking for a second opinion, which usually means another doctor. But a second opinion can also come from a second AI. I think of Claude as the second opinion, and ChatGPT, Grok, and Gemini as the third, fourth, and fifth. I have Claude export its “memory” of me and run it by the others. It takes fifteen minutes: no bills, no travel, no waiting rooms, no paperwork, and no insurance companies. The beauty of Claude is that I can program it to poll the other AIs on its own. This reduces the risk of hallucinations substantially.

Nothing I have described is perfect, but few things in life are. I am sure there is a risk of data leakage somewhere between Oura, my doctor’s portal, and the AIs. And every AI you share your health data with is one more place it could leak, so if you try this, turn off model training in each one’s privacy settings.

In the end, this is just a baby step toward taking greater control of my health. Notice something about both stories I opened with: they end with a doctor. My friend’s sight wasn’t saved by AI alone. It was saved by AI handing a doctor the right place to look.

That’s the model. Not AI instead of your doctor. Everyone is talking about AI becoming superintelligent. In health, at least, superintelligence is AI, HI (human intelligence), and MD combined. AI has become part of my medical team, and it turns me into the world’s best-prepared patient rolling into that Formula One pit stop.

I would love to hear your thoughts. How can I improve this? Please leave your feedback here.

P.S. Curious what a résumé of your body looks like? Here is a sample I made for a fictional patient, Joe Everyday Jr. Every name and number in it is invented, but it shows how the pages are laid out: data first, diagnoses next, questions last.

P.P.S. Health isn’t the only part of my life I run this way. I use Claude projects for taxes, personal finance, and house maintenance and remodeling.

P.P.P.S. Sleep, as I mentioned, has been my weakest link, so I have been working on its quality. Lately I have been experimenting with sleep pads. These are temperature-controlled mats that go on top of your mattress (think of the heated floors in your bathroom). They are heated and cooled by tiny water tubes and connected to a small unit about the size of a humidifier. To fall asleep, my body needs to cool down, so the pad cools the bed and I fall asleep faster. A few hours later my temperature keeps dropping, and that often woke me up. Now the pad warms the bed, and I don’t wake up as much.

I have tried two so far. Chilipad was a bit cheaper, but its motor was slightly louder (still tolerable), and the pad felt like I was sleeping on plastic. It had no sensors tracking my sleep and didn’t adjust to how I slept.

Eight Sleep (the Pod 6, the latest generation) is a substantial improvement. The pad feels like a normal mattress topper; you would not know there is an elaborate plumbing system underneath. The unit is so quiet you don’t know it is there. It has lots of sensors that track your sleep and adjust the temperature to your body. Eight Sleep is a significant step up, but it is more expensive than Chilipad, and it requires an annual subscription.

Both offer two zones, one for you and one for your significant other. I have a solo unit; my wife is not a fan. And to remove any ambiguity: I don’t get any kickbacks from either company.

Key takeaways

• Doctors are noble, overworked, and human. The visit is a Formula One pit stop, and in those few minutes your doctor has to reconstruct you from notes written months and five hundred patients ago. Life walks into the exam room with them.
• Your health is your own responsibility. Nobody cares about it more than you do, except maybe your mother — and of the seven factors that determine it (diet, exercise, sleep, stress, relationships, genetics, luck), five are in your control.
• Feed the AI everything: medical records, bloodwork, calendar, scale, Oura ring. Once a week is the right cadence. Daily is like checking your stocks daily — mostly noise, and it would make you obsessed.
• AI hallucinates, which is harmless when you ask for the capital of Morocco and not okay when it comes to your health. So don’t rely on one. Claude is the second opinion; ChatGPT, Grok, and Gemini are the third, fourth, and fifth. Fifteen minutes, no waiting rooms.
• Notice that both stories end with a doctor. The friend’s sight wasn’t saved by AI alone — it was saved by AI handing a doctor the right place to look. In health, superintelligence is AI, HI, and MD combined.