Research & methods
Transparent methods.
Honest limits.
Read how Birll calculates and presents HRV, Recovery, Sleep and Effort, with primary references and the limits of each method.
Reviewed 26 October 2026
Published research informs parts of Birll’s implementation. A reference to a method does not establish that the complete app, a daily index or an AI reply is clinically validated. Birll is for general health and wellness, for adults 18+.
01 / From method to measurement
How Birll uses each method.
HRV & signal quality
The metric and the source matter.
RMSSD describes successive beat-interval differences. SDNN describes the spread of normal beat intervals. They measure different properties and should not be treated as interchangeable. [1]
Birll’s Recovery path prefers available native Apple Health RMSSD and can fall back to legacy SDNN, keeping metric families and sources separate in the baseline. The HRV engine also implements artifact handling informed by published beat-classification work. Recording conditions and signal quality still affect the result. [2]
Recovery
A personal index, with explicit choices.
Recovery combines available HRV, resting heart rate, respiratory rate and wrist-temperature contributions against a rolling personal baseline. Available contributions receive equal weights. Sleep remains a separate pillar.
The displayed scale is an app-specific presentation of that index. It is not a probability of recovery, a diagnosis or a validated instruction about how hard to train. Missing data and changes of source affect the context available.
Effort
Recorded load, rather than readiness.
Effort summarizes recorded load using available heart-rate or perceived-exertion data. When those inputs do not provide usable load, it may use a labelled duration-and-energy estimate. The 0–100 display is an app-specific scale.
The age-based maximum-heart-rate path uses the Tanaka estimate, 208 − 0.7 × age. A group-based estimate is not an individually measured maximum. Effort and workload ratios are descriptive context, not injury predictions or a Recovery score. [3]
Sleep
Useful context, with measurement limits.
Birll summarizes available sleep records and stages from permitted sources. Its Sleep display uses app-specific choices; a stage estimate from a wearable is not the same measurement as a clinical sleep study.
Published work on wearable sleep technology describes device- and setting-dependent limitations. It does not establish the accuracy of Birll’s summaries or replace a clinical assessment. [4]
Patterns & correlations
An association is a starting point.
Birll explores relationships in the measurements available to it. Data coverage, uncertainty and the number of comparisons matter. A pattern is not a guarantee that changing a habit will change a score.
Two measurements moving together do not establish that one caused the other. Other influences, recording changes or chance may contribute. Research on causal language helps explain why this distinction matters. [5]
Lab results
Keep the laboratory’s context.
Lab reference intervals depend on the laboratory and its methods. Birll uses the interval recorded with your result rather than inventing a narrower “optimal” band. A reference interval is not a diagnosis or a treatment target. [6]
Report import currently captures the marker, value and unit. Add the reference range separately. An imported result without that range is not graded against one.
02 / Checks & limits
Checks and validation limits.
Engineering checks
The code includes synthetic known-answer fixtures, mathematical-identity checks and regression tests for selected calculation paths. Source review can verify a formula, the handling of missing data and how a display is produced.
These checks are useful evidence about the cases exercised. Their numerical tolerances are test assertions, not clinical accuracy specifications.
What remains unvalidated
Engineering checks do not establish prospective accuracy across people, test-retest reliability of the full workflow, diagnostic performance or health-outcome benefit.
Those questions require an appropriate study protocol, population, reference measurements and reported uncertainty. The papers below do not validate Birll as a complete product or establish superiority over another app.
03 / Primary publications & standards
Read the sources.
Each source supports the method or limitation named above. Its study population, recording conditions and scope remain important.
- Heart rate variability: standards of measurement, physiological interpretation and clinical use
ESC/NASPE Task Force. Circulation.
- A robust algorithm for heart rate variability time series artefact correction using novel beat classification
Lipponen JA, Tarvainen MP. Journal of Medical Engineering & Technology.
- Age-predicted maximal heart rate revisited
Tanaka H, Monahan KD, Seals DR. Journal of the American College of Cardiology.
- Wearable Sleep Technology in Clinical and Research Settings
de Zambotti and colleagues. Medicine & Science in Sports & Exercise.
- Causal and Associational Language in Observational Health Research: A Systematic Evaluation
Haber and colleagues. American Journal of Epidemiology.
- Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory
Clinical and Laboratory Standards Institute. Third edition, 2010.
Questions about a method?
Email info@birll.com. Please avoid sending health records or other sensitive details. Birll is not a medical device and does not replace advice from a qualified healthcare professional.