Data Dictionary
§ 7.0 — DATA DICTIONARY · EVERY VARIABLE, SOURCE, AND CAVEAT
If a number appears on this site, it is defined here: what it means, its unit, where it comes from, and how far to trust it.
| Source | What it provides | Reliability note |
|---|
| BodySpec DEXA (×2) | Body composition — total + regional fat/lean/bone mass, body-fat % | Gold-standard outcome, ±~0.5–1% per scan; lean includes water |
| Apple Health | Active/basal energy, heart rate, resting HR, HRV, sleep, VO₂max | Dense; device-dependent; protein sync is inflated |
| LoseIt food logs | Daily calories + protein | Canonical for protein & intake; only Date/Calories/Protein trusted |
| Strong | Per-set weight × reps by exercise | Logging starts 2026-04-16 (mid-window) |
| NHANES (public) | National DXA + demographics reference | Benchmark only; disabled offline |
| ACSM / BodySpec blog | Protein ranges; hydration coefficient | Published guidance / vendor experiment |
The analysis window is 2026-04-02 → 2026-05-02 (the two scans). Sources are most reliable inside it.
| Variable | Definition | Unit |
|---|
fat_pct | Body fat as a share of total mass | % |
lean_mass_lbs | Non-bone, non-fat tissue — includes water, blood, glycogen fluid | lb |
fat_mass_lbs | Fat tissue | lb |
lean_t0 / lean_t1 | Lean mass at scan 1 / scan 2 | lb |
lean_delta | lean_t1 − lean_t0 (per region or total) | lb |
lean_pct_change | 100 × lean_delta / lean_t0 — proportional change, the correct hypertrophy metric (absolute lbs favors large groups) | % |
lean_pct_band95 | the 95% precision band as a share of baseline (band95 / lean_t0) — wider for small regions | ± % |
fat_delta | fat_t1 − fat_t0 | lb |
fat_pct_change | 100 × fat_delta / fat_t0 | % |
lean_delta_band95 | 95% measurement band on the lean delta, from 1% per-scan CV in quadrature | ± lb |
lean_is_resolvable | abs(lean_delta) > lean_delta_band95 — does the change beat random noise? | bool |
lean_is_resolvable = true rules out random precision error only. It does not mean the change is muscle — systematic hydration error is a separate, larger effect (see the decomposition).
| Variable | Definition |
|---|
real_muscle_lbs | min(lean_delta, 4.29) — plausible newbie muscle, capped at 1 lb/wk × 30/7 |
water_lbs | lean_delta − real_muscle_lbs — remainder attributed to glycogen/hydration |
water_gallons | water_lbs / 7.4 — water-equivalent, via BodySpec's gallon experiment |
water_pct | water_lbs / lean_delta — share of the "gain" that is likely not muscle |
| Variable | Definition | Unit |
|---|
active_kcal | Active energy burned (movement) | kcal |
basal_kcal | Basal/resting energy | kcal |
tdee_kcal | active_kcal + basal_kcal | kcal |
intake_kcal | Food calories from LoseIt | kcal |
net_balance_kcal | intake_kcal − tdee_kcal (negative = deficit) | kcal |
cumulative_net_kcal | Sum of net balance over logged window days | kcal |
implied_fat_change_lbs | cumulative_net_kcal / 3500 | lb |
observed_fat_delta_lbs | DEXA-measured total fat change | lb |
fat_reconciliation_gap_lbs | observed − implied | lb |
protein_g | Daily protein (LoseIt food logs) | g |
protein_g_per_kg | protein_g / (weight_lbs / 2.20462) | g/kg |
ah_protein_g | Apple Health protein — cross-check only, ~2× inflated | g |
| Variable | Definition | Unit |
|---|
volume_lbs | Σ (weight × reps) over working sets, by region/day | lb |
working_sets | Count of non-warm-up sets | n |
region | DEXA region an exercise maps to (arms/legs/trunk) — shoulders → arms | — |
resting_hr_bpm | Daily mean resting heart rate | bpm |
hrv_sdnn_ms | Heart-rate variability, SDNN | ms |
sleep_hours | Time asleep (summed asleep segments) | h |
| Variable | Definition |
|---|
pct20_bf … pct80_bf | Body-fat % cut points by sex + age band; e.g. for men 20–29: 16 / 20 / 24 / 27. A value below pct20_bf is leaner than ~80% of the group |
dim_acsm_targets | Reference ranges — protein 1.6–2.2 g/kg (hypertrophy), volume guidance |
- DEXA lean = tissue + water. The single largest confound; the whole centerpiece exists to address it.
- Window/logging mismatch. Strong logging covers ~11 of 31 window days, so regional volume understates true training.
- Small n. Two DEXA endpoints and three regions — correlations are illustrative, not significant.
- Model approximations. 3,500 kcal/lb and a flat 1% CV are first-order assumptions, exposed as tunable parameters, not ground truth.
- Single subject. Everything here describes one person over one month; nothing generalizes.
↳ how these are computed: [methods](/methods) · why it's framed this way: [about](/about)
