A displayed value is not always a new observation
Source and acquisition date are essential metadata for longitudinal interpretation. A carried Cardiometabolic or Functional value may support a model calculation without establishing that a new measurement occurred on the displayed day. VO₂max, Walking Speed, and Cardio Recovery retain their original dates; their age lowers Data Confidence linearly across 30 days without changing modeled ΔYears. Daily Behavioral exposures, HRV, resting heart rate, and sleep remain on their own daily records.
An episodic laboratory clock has a valid panel date and recency history of its own. Carry-forward conventions cannot make its underlying laboratory assessment daily.
Support and recency answer different questions
BioMIR uses Data Confidence to describe support for the daily ABA estimate; age-based support applies to carried Functional as well as Cardiometabolic measurements. Separate Data Freshness displays track Cardiometabolic anchors and clinical panels. These are data-support indicators, not numerical probabilities that an age estimate is correct. The Functional 30-day window is an operational convention, not a biological half-life or clinical expiration.
Missingness, late-arriving records, device updates and changed recording conditions can produce apparent shifts or discontinuities unrelated to physiological change.
Measurement conditions are part of provenance
Resting heart rate, HRV and wearable-estimated sleep or fitness measures depend on sensor method, activity state, algorithm and acquisition conditions. Comparisons are strongest when source and conditions remain reasonably consistent.
Consumer wearable performance varies by metric and device. Device-level validation is therefore not interchangeable with validation of a final biological-age composite.
Practical longitudinal checks
- Inspect the original observation date rather than relying only on the chart date.
- Compare matched periods, units and statistic settings before interpreting a change.
- Check whether new, missing or carried input streams explain a shift in the modeled output.
- Distinguish data availability and recency from biological stability or clinical certainty.
Measurement error and observation-process bias
Data quantity and data validity are separate. A dense time series produced by a sensor with drifting algorithms can give a false impression of precision, while sparsely collected but carefully standardized laboratory observations may supply a different form of evidence. Changes in device generation, software version, sampling protocol or user behavior can create discontinuities that resemble physiology.
For rigorous comparisons, maintain source identifiers, original collection timestamps, units, transformations and the reason for retaining or excluding each observation. Evaluate whether missingness correlates with illness, activity, travel or intervention engagement. The observation process itself can be an important source of selection bias.