The short arc, told without receipts.
Late 2023, early 2024 I was in the best shape I'd ever been in. Lifting heavy, recovering fast, the Apple Watch estimating my VO₂max around 55. Then I went to Berkeley on a semester exchange, and somewhere in the middle of it the stress load exceeded what my nervous system could buffer. It started as dizziness. It became dizziness severe enough to put me in the ER getting checked for a cardiac event. The workup came back clean. What was actually happening was a panic disorder on a fuse. By Q1 2025, back in Germany, a therapist gave it a name.
What I didn't know until later is that a nervous system in alarm mode is also a gut on fire. My February 2025 YEARS annual captured both at once — calprotectin at 434 µg/g, VO₂max collapsed below 33, HRV percentile ranks near the floor. A colonoscopy turned up early signs of Crohn's. I was 86 kg at that point, 12 kg lighter than four months earlier, most of it muscle. You don't lose muscle at that speed unless something is wrong at the system level.
What came next — the gut protocol, the therapy, the slow return to jogging and then to lifting — is the content of the H1 2026 drop. April 2026 I'm 95 kg, VO₂max 42.0 at 98% predicted, calprotectin 9, HRV climbing on every axis, panic gone. The story isn't over. It isn't supposed to be — the point of Peakspan is the span of time at peak, not a single peak moment.
2024 was survive and be healthy again.
2025 was rebuild the foundation.
2026 is reach peak — and stay there.
Why open-source this.
- 01
Because the data stops being theoretical when it's yours.
Every number in this experiment sits in some clinical study as a mean with error bars. Here, one of those error bars has a name. That's not more informative than a meta-analysis — it's differently informative. Specifically: it's an existence proof. What happens to a single body under a specific stack, across a specific window.
- 02
Because the bad data gets posted, not buried.
Most online health content has selection bias baked in — people talk about what worked. I'm committing to publishing the things that didn't move, the interventions I walked back, and the labs that drifted the wrong way. If a drop is boring because nothing changed, that's still the drop.
- 03
Because it keeps me accountable.
There's a quiet selfishness to this: public data makes it harder to skip the uncomfortable retest. I already know which markers I don't want to look at this cycle. That's exactly why they're on the page.
- 04
Because maybe it's motivating, or useful.
A reader might spot a pattern. A clinician might see a line to pull. Someone else with an Lp(a) number they can't reason with might find this more digestible than a 40-page guideline. Or none of the above. The site exists either way.
What Peakspan is — and isn't.
This site is
- One person's open biomarker ledger
- A half-yearly, cited, date-stamped record
- A methodology page you can argue with
- Honest about what didn't work
- Free, ad-free, sponsor-free, affiliate-free
This site isn't
- A protocol to copy
- Medical advice
- A supplement store
- Generalisable to any other body
- A replacement for a clinician who knows you
If you want the rest of me.
Peakspan is just the biomarker ledger. Everything else I do lives elsewhere — the writing, the podcast, the work at YEARS, the occasional speaking. If that's what you came for, the right door is nikohems.de.
Head of Growth at YEARS, a preventive-medicine clinic in Berlin. Host of the Return on Health podcast. Writes essays at nikohems.de. Does not sell supplements, does not do affiliate, has no financial relationship with any of the labs, providers, or products named on this site beyond being their patient and / or paying customer.