Precisely measured is not the same as useful.

The market divides into four types. Private preventive centres bundle laboratory, imaging and functional diagnostics into one day; these include YEARS in Berlin and Metamedicum in Essen and Düsseldorf, formerly Preventicum. University centres such as the human genetics institutes deliver the deepest molecular diagnostics but work from a specific clinical question rather than as screening without indication. Medical-spa houses such as Lanserhof combine diagnostics with a multi-day stay. Providers using partner clinics such as Aeon cover several cities. The choice depends less on equipment than on whether medical interpretation follows the report and whether evidential weight is declared openly per procedure.

The attempt to tailor prevention to the individual rather than assign it by age and sex. Classical prevention works with group rules: colonoscopy from 50, mammography from 50, health check-up from 35. Precision prevention instead aims to estimate individual risk and derive intervals and procedures from it. It rests on three components: risk stratification from family history and clinical scores; molecular markers from lipoprotein(a) through inflammatory markers to genetic findings; and imaging as a structural snapshot. The term is not protected and describes no defined scope of service.

Partly, and the differences are large. Well evidenced is risk stratification with established markers: measuring Lp(a) once in a lifetime, ApoB rather than LDL alone, confirming familial hypercholesterolaemia by genetic testing, screening Lynch syndrome carriers earlier and more often. These examples have guideline status. Weakly evidenced is the leap from a broad marker or imaging package to a better course. For whole-body MRI without indication, multi-cancer blood tests and polygenic risk scores, randomised trials with patient-relevant endpoints are missing. The procedures measure reliably. That the measurement ultimately leads to more healthy years has not been shown for this group.

By a chain with five levels, and most discussions skip the middle. First: the value is reliably measured. Second: the test detects a condition. Third: the finding survives the work-up and would not otherwise have surfaced. Fourth: the diagnosis leads to a treatment decision that would otherwise not have happened. Fifth: that decision changes the course. Only the fifth level is the benefit; everything below is a precondition. The most common error in marketing preventive diagnostics lies not in the choice of procedure but in the silent leap from the first level to the fifth.

Worth taking seriously as a research instrument, not yet as a basis for decisions. A polygenic score combines hundreds to thousands of variants, each with a tiny effect, into a risk estimate. Three limitations. First, the score shifts a probability without establishing a diagnosis. Second, it is calibrated to the cohorts in which it was developed, so predictive performance outside European populations is far less well documented. Third, for most common conditions it explains only part of the variance, while blood pressure, ApoB, HbA1c, smoking and exercise explain the larger part. A high score rarely changes the recommendation; it can shift priorities.

We use the procedures and assign each to one of three evidence tiers, publicly and per procedure. Tier one is guideline- and endpoint-based, including lipid diagnostics with ApoB and Lp(a), monogenic genetic findings and pharmacogenetics. Tier two is study-based without guideline status. Tier three is explicitly research-adjacent: whole-body MRI as screening, liquid biopsy, epigenetic clocks, microbiome analysis. For tier three we state that no diagnosis and no treatment follows from it. We use it because the longitudinal data we want to build does not yet exist — not because it is already evidenced. That separation is set out in full on our evidence page.