I track roughly 120 biomarkers across blood panels, metabolic markers, hormones, lipids, and inflammation indicators. I test annually through Function Health (or similar) and maintain a spreadsheet where each marker has an optimal range, an importance weighting, and a score representing how close my last result is to the midpoint of that range. For things outside of the range, or drifting towards the limit over time, I research what intervention has strong evidence behind it. As with most things, Claude does a good job finding applicable. research

I’m not a doctor. This page only covers how I think about supplements and health monitoring. This is what I do for myself, after reading more papers and blog posts about it than is probably healthy.

I admittedly don’t exercise enough. It has a better return than anything on this page, but I find it hard to make time.

How I choose supplements

I apply three filters before adding anything to my routine:

Long history of use without serious adverse signals. If a compound has been consumed by large populations for decades (creatine, fish oil, vitamin D, fiber) and the safety literature is has no red flags. I avoid novel compounds.

Research verifying improvement. I want to see either a meta-analysis, a well-powered RCT, or at minimum a consistent direction across multiple independent trials. A proposed mechanism of action isn’t enough… most tested compounds don’t actually work. Examine.com is good for quickly evaluating the evidence behind any supplement. The free version used to be good; now it’s $29/month but I tend to only buy it for a session.

Reliable and tested source. Supplements aren’t regulated well. I look for third-party testing and generally buy from brands that publish certificates of analysis.

Why I don’t take a multivitamin

Most multivitamins contain compounds that don’t pass my requirements above, or with doses too low or too high. Beyond a few things that the Western lifestyle cause in nearly everyone, supplement needs are individual. Annual bloodwork tells me what I actually need. A multivitamin is like using a shotgun without aiming.

Supplements I think most people should consider

A supplements seem to pop up every time I do broad research and seem to be near universal.

Vitamin D

Unless you work outside for hours every day, you are likely deficient. Vitamin D is cheap, safe at reasonable doses, and the evidence base for lowering all-cause mortality is large. Gwern has a good writeup. Most people in northern latitudes should be taking 2,000–5,000 IU daily, though blood work is always better then guessing.

Creatine

Creatine monohydrate is famous among bodybuilders, but the cognitive benefits are underappreciated. Gwern’s meta-analysis of creatine and cognition covers this well. 5 g/day of creatine monohydrate is the standard dose. It is cheap, extremely well-studied, and safe.

Creatine will increase serum creatine level, which can make eGFR (kidney function estimate) look worse on a blood test. If worried, there are other measurements feasible to check kidney funciton.

Metformin or GLP-1 agonists

Metformin is one of the most interesting longevity candidates. It’s an old, cheap diabetes drug with a broad evidence base for reducing all-cause mortality, cancer incidence, and cardiovascular events.

GLP-1 receptor agonists (semaglutide and similar) are newer but showing remarkable effects beyond weight loss, including cardiovascular risk reduction. Semaglutide was only approved in 2017… which is right on the edge of my ’long history'.

Both require a prescription in the US, but other countries exist.

Coffee

I’m addicted, so take this with a grain of salt and some motivated reasoning. That said, the evidence is strong. A large study in Circulation found that moderate coffee consumption (roughly two cups per day) is associated with significant risk reductions for cardiovascular disease, heart disease, and stroke. There are also the obvious day-to-day benefits of improved alertness and mood that I would struggle to give up even if the long-term evidence were neutral.

Fiber

Most Westerners eat about half the recommended daily fiber intake. Fiber lowers LDL cholesterol, improves both constipation and diarrhea, and slows glucose absorption.

I supplement with oat β-glucan and psyllium husk. β-glucan for the microbiome and cholesterol, and psyllium for bowel regulation and glucose control.

Drink plenty of water.

For colds: zinc and vitamin C

When I feel a cold coming on, I take zinc lozenges. There is evidence that zinc reduces the duration and severity of colds. Brand and formulation matter. Zinc acetate lozenges seem to work best, and many commercial products use doses or forms that don’t match what worked in trials.

I also take vitamin C for a cold. The evidence is weaker though there may be a small reduction in duration. But it’s cheap and safe… so why not.

How I analyze my metrics

I maintain a spreadsheet of every biomarker I’ve tested, going back several years. For each marker I track:

• Historical values across test dates, so I can see trends rather than reacting to a single snapshot.
• An optimal range, not just the lab reference range. Lab ranges tell you “not obviously diseased.” Optimal ranges are “where you want to be.” These come from a mix of Function Health’s recommendations, primary literature, and longevity-focused blogs (mainly Peter Attia and Bryan Johnson). I’m careful to note where my “optimal” range comes from and how confident I am in it. Most of my ‘optimal’ ranges are tighter than the lab’s and have an ‘ideal’ measurement which is usually the middle of the range, or zero for measurements where the goal is no detection. I only use an ‘optimal’ outside the lab measurement if there is research showing it is not harmful.
• An importance weighting (1–10) reflecting how much the marker matters for my specific risk profile. I used an LLM to generate these and then sorted to verify.
• A score that quantifies how close my last result is to the ideal measurement. Each biomarker gets a score from 0 (far outside optimal) to 1 (at the ideal measurement), with a steeper penalty once the value leaves the optimal range:

When a marker is out of range, I research interventions in roughly this order: dietary changes first, then supplements, then medication.

What I watch most closely

My genetics (via 23andMe) inform what I prioritize. I won’t go into specifics, but if you know your genetic risk factors and family history, weigh your attention. A marker that’s borderline normal in someone with no genetic risk is an issue for someone with a genetic predisposition.

Beyond genetics, I pay closest attention to markers with high importance weightings: cardiovascular risk (ApoB, LDL particle number, Lp(a), hs-CRP), metabolic health (HbA1c, fasting insulin, glucose), hormones, and inflammation markers.

The meta-process

Perhaps the most useful thing I’ve learned is to treat health as an empirical question:

• Assume your priors are wrong. Bloodwork is the arbiter, not subjective experience.
• Beware of motivated reasoning. I want my supplement stack to be working. That makes me a bad judge of whether it is.
• Retest. An intervention not retested does not provide evidence.
• Read the counterarguments. For every supplement with positive trials, find the null results too. If I can’t steelman the case against it, I haven’t done enough research.

The goal is to find the handful of interventions with the best ratio of evidence strength to cost and effort, implement them consistently, and verify they’re actually working.

Results

As of 2026-03-15, my overall weighted score has improved 2% since starting. Considering I’m aging, I consider any improvement as good.

If I use the typical lab ranges, my weighted score is 70%. 35% is the edge of the bound and 100% is the ideal… so 70% is, on average, in the range but half way to one of the limits.