I have a Withings scale that has weighed me almost every morning since August 2014. An Oura ring that has slept with me every night since 2020. An Apple Watch on my wrist since the day Apple started making them. A Withings blood pressure cuff in the bathroom drawer. A continuous glucose monitor patch on my arm right now, transmitting interstitial sugar every five minutes. Sixteen full blood panels in a folder. A genomic report from Bioma covering forty-seven metabolic variants. A DEXA scan from BodySpec that knows the exact pound of visceral fat sitting against my liver.

For most of those eleven years, I did nothing with any of it. I’d glance at the apps when they pinged. I’d notice a weight number on the scale. I’d see “your sleep was 78” without having any idea what that meant. The data accumulated in the background like email I never opened. I knew it was there. I assumed someone, at some point, would tell me if it mattered.

Nobody ever did.

The trove

If you also own wearables, this list will sound familiar. Maybe even small. Here is what was sitting on my devices when I finally started looking:

I am not unusual. Anyone reading this who has worn an Apple Watch for five years and owned a smart scale and gotten labs done at a physical is sitting on roughly the same volume of personal biometric data. The difference between me and you is not the data. It is the three months.

The blind spot

Here is the moment the project actually started. In February 2026 I uploaded my Oura ring data — the full export, every night going back to 2020 — to a conversation with an AI. I asked a vague question. I don’t remember what. Something like “is there anything weird in here.” I expected nothing. Wearables tell you “your sleep is good” or “you should move more.” That’s what they always said.

What came back was a chart that I had never seen before. My resting heart rate from late 2020 to early 2026. Five and a half years of overnight averages, drawn on a single axis. And there it was, plain as a balance-sheet trend line: my resting heart rate had been climbing for almost three years.

Forty-nine point five in November 2022. Fifty-four by mid-2023. Fifty-eight by mid-2024. Sixty-two by mid-2025. Seventy-two by April 2026. A 23-bpm climb in three and a half years, with nothing on my dashboards that had ever flagged it. Every individual night looked “fine.” Each week’s average was a little higher than the last, but only by a beat or two — under any normal monthly noise floor. The pattern was invisible at any timeframe shorter than years.

The same AI that drew the chart suggested obstructive sleep apnea. Resting heart rate climbing during sleep, with no corresponding fitness decline, is one of OSA’s quiet signatures. I had been diagnosed with mild OSA in 2024 and given a CPAP machine. I had been inconsistent with it. The climb tracked, almost perfectly, my CPAP non-compliance. The ring had been recording the consequence in real time. Nobody had told it what the data meant.

That was the first one.

The thyroid arc

The second pattern showed up in the blood panel folder. Sixteen panels over five years. Each one had been reviewed by a doctor at the time it was drawn. Each one had been declared “within range.” That is — every individual reading sat between the lab’s published reference floor and ceiling. So no flag was ever raised.

What no individual visit captured: the trajectory.

My Free T4 — the active thyroid hormone — had been drifting downward. From 0.98 ng/dL in 2021 to 0.70 ng/dL in March 2026. Each value still “normal.” But the line, when you actually drew the line, sloped down by about twenty percent over five years. With TSH staying flat at 1.23. A normal TSH should be telling the thyroid “you’re doing fine.” Either the pituitary isn’t signaling correctly (central hypothyroidism), or the conversion from T4 to T3 is failing somewhere downstream. Either way, this is a question. Five years of single-visit “all normal” never raised it.

What the labs missed

This is the structural problem with how clinical data is currently presented to patients. Lab reports are point-in-time. They compare your value to a population reference range. They do not compare your value to your own value last year, or your value the year before that. The trajectory information is in the data — your portal has every result going back as far as you’ve been a patient — but nobody is drawing the line.

Drawing the line is not hard. It’s a one-liner in Python. It’s a prompt to any modern language model. The reason it doesn’t happen at scale is that a primary care doctor has fifteen minutes per visit and three hundred patients on their panel. They are not running longitudinal regressions on Free T4 in their off hours.

The interpretation layer

This is what the experiment is really about. The thing I keep saying, but it bears repeating clean: the data was already there. The Oura data was already there. The blood panel data was already there. The Withings data was already there. None of it was a discovery. The discovery was that nobody had ever looked at it sideways.

For somebody like Bryan Johnson, this is solved by hiring a team. A reported team of thirty plus people, on a reported budget north of two million dollars a year, whose actual job is to read his biometric data and tell him what it means. That’s the interpretation layer, staffed.

For me — and for everyone I know who owns wearables — the interpretation layer has been missing entirely. The data accumulated. The interpretation didn’t.

What’s changed in 2025 and 2026 is that the interpretation layer became accessible. Not perfect, not at the level of an endocrinologist looking at your charts in person — but real. I can sit on the couch and upload three years of overnight SpO2 data and have a working hypothesis about my sleep apnea ten minutes later. I can paste sixteen blood panel rows and have a downward trajectory of Free T4 charted, the differential diagnosis between central hypothyroidism and conversion failure framed in language I can take to my endocrinologist, before lunch.

This used to cost two million dollars and require a staff. This costs the price of a coffee subscription and requires you to know how to paste a CSV.

What this experiment is really about

If you’ve stuck with me this far you might be expecting a supplement stack, a protocol, a recommendation. That is in the rest of the journal and on the dashboard. It’s not the point of this post.

The point of this post is the precondition. Before any protocol, before any intervention, before you spend a single dollar on a supplement or a peptide or a continuous-glucose monitor you don’t need, the leverage is in the data you already have. Almost everyone in the demographic I am writing for — men in their forties and fifties, professional, mildly metabolically nervous — already owns the devices. The watch is on the wrist. The scale is in the bathroom. The lab portal has a decade of results.

The intervention I would recommend before any other intervention is: read your own data, longitudinally, at least once a year, with someone (or something) capable of seeing trajectories. Not your annual physical, which is structurally incapable of doing this. Not your wearable’s auto-generated weekly summary, which is structurally designed not to alarm you. Look at the line that connects 2020 to today. Ask whether anything is sloping somewhere it shouldn’t.

For me, that exercise — three months in, still not done — has produced two undiagnosed clinical findings (rising resting heart rate, declining Free T4) that I am now actively working on with two different physicians. It has produced a third (genomic NAFLD risk at 43 percent) that has changed what I eat for dinner. It has produced a working baseline for the next ten years.

It has cost approximately the price of a few months of a chat subscription.

The trove was already there. The interpretation is the experiment.