Beyond the annual physical

Conventional preventive care is episodic by design: a yearly visit, a narrow panel, a comparison against a population reference range. A result inside that range is called normal even when it represents a marked change from a person's own prior values.

Longitudinal measurement addresses that specific weakness. Repeated values create a personal baseline, and deviation from a personal baseline is a different and often more informative signal than deviation from a population mean. Consumer platforms such as Function Health have built businesses on this premise, offering broad panels with repeat testing and consumer-facing interpretation.

That is an interesting model, and it is not the same thing as demonstrated clinical benefit. Company claims and independent evidence are separate categories, and they should be evaluated separately.

The Cochrane review of general health checks in adults found that broad, routine health checks did not reduce all-cause, cardiovascular or cancer mortality, while increasing the number of new diagnoses. More looking produced more findings without producing more survival.

USPSTF evidence reviews across screening topics describe the mechanisms behind that result: false positives, overdiagnosis of conditions that would never have caused harm, diagnostic cascades of follow-on imaging and biopsy, and the anxiety, cost and procedural risk those cascades carry.

Peer-reviewed appraisals of direct-to-consumer laboratory testing describe a similar tension. Access and engagement can genuinely improve; so can misinterpretation, unnecessary follow-up and inequity when out-of-pocket testing sits outside a care relationship.

More data is not better healthcare. More data is more data.

The useful version of precision health is narrower and harder than the marketed version. It requires that each test be validated for the purpose it is used for, that results be interpreted against a person's clinical history, and that an abnormal value connect to a defined action.

• The right information — validated measures, chosen for a reason.
• For the right person — risk-stratified, not universal.
• At the right time — intervals that reflect biology, not billing cycles.
• Interpreted in the right clinical context — by someone who knows the history.
• With a clear action plan — including the plan to do nothing, stated deliberately.

Patients are already arriving with panels their organizations did not order and cannot bill for. Someone has to decide who reviews that data, what the documentation standard is, how incidental findings are escalated and what the organization will not act on.

That is not a technology decision. It is an operational and medico-legal one, and in most organizations it has not been made.

Longitudinal, personalized data is a credible direction for prevention and a poor substitute for evidence-based screening. Both statements are true at once, and holding them together is the discipline this field currently lacks.

The measure I would apply to any new panel, sensor or platform is unchanged from the one I apply to any operational investment: does it change a decision, and is someone accountable for the decision it changes?