Abstract
Estimates of cardiorespiratory fitness metrics (e.g., maximal oxygen uptake) from wearable electronic devices are often characterized by significant noise and variability. Inconsistencies arise due to algorithmic differences between manufacturers, sensor inaccuracies, and biological variance. The integration of data from multiple disparate sources lacks a standardized method for weighting measurements based on source reliability or handling late-arriving data. While the disclosed method is described in the context of cardiorespiratory fitness metrics such as VO2 max, it is equally applicable to any changing physiological variable derived from noisy or disparate sources, including but not limited to daily or weekly average blood pressure.
A method is disclosed wherein a Kalman filter is utilized to fuse measurements from multiple first-party and third-party sources into a single estimate. Operation is performed through a dual-trigger mechanism: daily triggers update confidence levels by adding process noise, while event-based triggers refine the state estimate using new measurements. Late-arriving data is incorporated through recursive recalculation restricted to current and future values. Adaptive expiration thresholds are established by monitoring the variance output of the filter. This approach provides a unified metric with quantified confidence, improving the reliability and consistency of health data presentation.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Phillips, Justin; Roggen, Daniel; and Harle, Robert, "Fusing Fitness and Health Sensor Data from Multiple Sources", Technical Disclosure Commons, (February 02, 2026)
https://www.tdcommons.org/dpubs_series/9254