13 posts
A direct look at blood-chemistry clocks, methylation clocks, and plasma proteins: which markers carry the signal, how to compute biological age from your own lab data, and how much of the number is noise.
A fitness or gym blood panel is only interpretable if you control what you did in the 72 hours before the draw. Here is the panel we would order, the confounders that corrupt it, and how to analyze the results over time.
Morning, fasted, rested, and identical from draw to draw. Here is which analytes move across the day, what the 48 hours beforehand do to your numbers, and how to write a draw protocol you can repeat for years.
A working setup for longitudinal biomarker tracking: a schema that survives lab changes, your own analytical and biological variation estimates, reference change values instead of reference ranges, and trend fitting that does not fire on noise.
What a blood biomarker test measures, what it costs, how to get the numbers out in a form you can compute on, and the math that tells you whether a change between two draws is real.
A working list of blood biomarkers that change decisions, why panel size is a poor proxy for value, and how to handle biological variation so you don't chase noise.
The 56-biomarker performance panels sold as a full-body MOT test the wrong thing in the wrong way. Here is the panel we would order, how to control preanalytics, and how many repeats you need before a number means anything.
What annual bloodwork is worth drawing, why vendor 'optimal ranges' are weaker than they look, and how to build a personal baseline using biological variation and reference change values.
The best blood test tracker is a long-format table you control plus fifty lines of Python, not an app. Here is the schema, the LOINC and unit normalization, the reference change value math, and the failure modes that make most trackers wrong.
A practical protocol for estimating functional age from physical performance tests, blood chemistry algorithms, and DNA methylation clocks, with the reliability problems that make most single-timepoint results uninterpretable.
A working pipeline for taking Olink, SomaScan, or DIA-NN output into R: QC, missingness, normalization, and longitudinal within-person modeling of a single individual's plasma proteome.
A practical comparison of DNA methylation tests: which array, which clocks, how to get the raw IDATs, and how to process them yourself with sesame and dnaMethyAge. Plus the noise sources that make a single epigenetic age number close to meaningless.
There is no single 'Huberman blood test.' Here is the panel that gets discussed on the podcast, which markers carry real signal, and how to get the data in a form you can analyze yourself.