About · Clinical references
The published sources behind every range and score.
BioElle's scoring, reference ranges, and clinical thresholds are grounded in published, peer-reviewed sources. Nine evidence domains are summarised below, each citing its primary references.
Lean mass, fat mass, visceral fat, regional distribution.
In short: we compare your scan with large studies of people your age and sex. These are guides, not a diagnosis.
Reference ranges are anchored to NHANES-derived percentiles, with sex-specific distributions and age stratification. Lean-mass and fat-mass context uses the percentile reference standards published by Imboden et al. (2017) — two companion papers, one for fat measures and one for lean measures. Both were measured on GE Lunar DXA in a Caucasian US sample, and the authors caution that the values “may not be an accurate representation for all ethnic groups”. They are reference standards, not diagnostic thresholds, and they do not transfer cleanly to Hologic scanners or to home BIA. Visceral adipose fat area is read directly from Hologic and GE Lunar scans; scans from other machines are read with a step where you confirm the values first.
Optimal vs. flat ranges, ferritin / iron, lipid panels, thyroid.
In short: a normal range fits 95 in 100 healthy people. A result just outside it is a reason to look closer, not a problem in itself.
A clinical reference interval is the central 95% of values — the 2.5th to 97.5th percentile — from a screened healthy reference population, partitioned by sex and age (CLSI EP28-A3c). By construction, 5% of healthy people fall outside it, so a result outside the range is a prompt to look closer, not a finding in itself. Alongside these, BioElle shows narrower "optimal" bands drawn from the research literature. These are not clinical thresholds, are not endorsed by any guideline body, and are not diagnostic. Ferritin deficiency thresholds follow the WHO 2020 guideline (below 15 µg/L in non-pregnant women; below 70 µg/L where inflammation is present, which requires inflammatory markers to interpret). WHO does not define an upper limit, and UK clinical practice commonly uses a higher threshold for iron deficiency. BioElle covers 34 biomarker types.
Evidence-grade dosing windows, interaction warnings.
In short: we don't recommend supplements. We track what you already take and show whether it lines up with your results.
Supplement protocols use Examine.com-style evidence grading (Strong / Moderate / Weak) and surface interaction warnings between common stack combinations. BioElle does not recommend supplements; it organises what you are already taking and shows whether it correlates with outcomes.
Phase boundaries, hormonal patterns, fluid retention.
In short: water can move your weight by about 0.3–0.5 kg across your cycle, so we count the scale less in those days. Your body fat itself doesn't change with your cycle.
Cycle phase resolver follows standard 28-day model adjusted for personal variability. Follicular, ovulatory, mid-luteal, and late-luteal boundaries shift estradiol and progesterone reference ranges accordingly. Cycle-related fluid shift is the basis of the volatility filter. Repeated-measures studies put the group-mean body-mass excursion at roughly 0.3–0.5 kg, driven by extracellular water and peaking around menstruation (Kanellakis et al., 2023; Gleichauf & Roe, 1989). Body composition itself — fat mass, lean mass, body fat percentage — does not show a reproducible shift across the cycle (Ong et al., 2022; Kuikman et al., 2024; Hicks et al., 2017), so the filter down-weights scale weight rather than adjusting composition.
Protein thresholds for active women.
In short: active women are usually advised 1.6–2.2 g of protein per kg of body weight a day.
Protein guidance uses 1.6–2.2 g/kg body weight per day for active women, the range established by Morton et al. (2018) and carried in the ISSN and ACSM position stands. Protein guidance is surfaced contextually, not as a verdict.
Phase-aware training, strength capacity, HRV.
In short: your cycle phase barely changes how you perform, so we don't change training advice by phase. Recovery scores only shift in the late luteal phase, the days before your period.
Cycle phase has only a trivial, low-certainty effect on exercise performance: the largest meta-analysis found a pooled effect of −0.06 across 78 studies, with no significant difference between strength and endurance outcomes (McNulty et al., 2020). BioElle therefore does not adjust training guidance by phase. The one cycle effect on heart-rate variability that survives meta-analysis is a reduction in the late luteal phase (Schmalenberger et al., 2019), so that is the only phase in which readiness thresholds shift. Recovery windows are tracked separately so training pauses don't misread as detraining.
Sleep debt, recovery thresholds.
In short: we use how long you slept, compared with your usual. We don't score deep or REM sleep, because watches and rings measure them poorly.
Sleep is read as total nightly duration from Apple Health, compared against your own rolling 7-day baseline with the two most recent nights excluded, so a single short night doesn't reset the line it is being judged against. BioElle deliberately does not score individual sleep stages. Validation studies against polysomnography show consumer wearables estimate deep sleep with per-night errors of the same order as the quantity being measured, and the direction of that error is not consistent between devices — so a stage-level threshold would read as far more precise than the underlying data supports.
How patterns are detected — and when we hold back for lack of evidence.
In short: we only show a pattern when there's enough data behind it.
Behavioural patterns are only surfaced when there's enough data behind them — a minimum number of readings — so a one-off blip never becomes an "insight." Predictions always show how confident the estimate is. The methods are deliberately conservative: we'd rather show nothing than a shaky pattern.
Glycaemic, thyroid, iron, cycle & androgen, heavy-bleeding, and post-surgical marker patterns.
In short: when several results point the same way, we name the conditions studied alongside that pattern, so you can ask your doctor a clear question. We never say you have them.
When enough markers point the same way, BioElle surfaces the pattern and names the conditions it is studied alongside at the population level — so you can raise a specific, evidence-anchored question with your doctor. Glycaemic patterns are studied alongside pre-diabetes and type 2 diabetes (ADA Standards of Care in Diabetes—2026; NICE PH38, which covers prevention and high-risk identification rather than diagnosis). Where insulin resistance is named, it is named without a threshold: no guideline body defines a diagnostic cut-off for it, and HOMA-IR is not endorsed for individual assessment. Thyroid-axis patterns are associated with subclinical hypothyroidism and autoimmune thyroiditis (NICE NG145). Iron-status patterns track with iron deficiency (NICE CKS, Anaemia – iron deficiency; British Society of Gastroenterology, Snook et al. 2021). Cycle-and-androgen patterns are studied alongside PCOS and congenital adrenal hyperplasia (Teede et al. 2023, which covers both). Heavy-bleeding patterns are investigated alongside fibroids, adenomyosis, and endometrial pathology (NICE NG88 for all three; RCOG Green-top 67 for the endometrial limb only — its scope is endometrial hyperplasia). Post-surgical patterns reflect nutrient-deficiency risks after bariatric procedures (BOMSS / ASMBS micronutrient guidance).
BioElle names these conditions at the population level only. It describes trends seen in people with similar markers — it never tells you that you have a condition, and always points the conversation back to your GP or doctor.
BioElle suggests patterns. It does not diagnose. Always close the loop with your GP or doctor. Full evidence sources are available on request.