About Nutrinaut Tools

Methodology and Sources

Source hierarchy, formula registry, validation rules, product heuristics and version history for every Nutrinaut tool.

Transparent, deterministic and versioned

Calculator outputs are deterministic local calculations, not AI diagnoses. The same valid inputs produce the same result. Metric values are the canonical calculation path; pounds convert using 1 lb = 0.45359237 kg, feet using 30.48 cm and inches using 2.54 cm.

References support the method or context; they do not validate an individual result. Every numerical input is checked for a finite value inside a defensive product bound. Invalid values are rejected rather than silently clamped.

How evidence is selected

  1. Official standards and regulators. FDA, EFSA, EU regulations, USDA, NIH and other responsible public bodies.
  2. Professional consensus and systematic review. Used when the scope matches our recreational adult audience.
  3. Primary research. Used to establish an equation or clarify uncertainty, not to overgeneralise a single study.
  4. Product heuristics. Clearly labelled interface rules used for display or planning.

Eight core methods and practical comparison labs

ToolCore logicCanonical unitsDisplay roundingVersion
Maintenance & Energy NeedsMifflin–St Jeor REE; × separate FAO PAL rangekg, cm, yearsNearest kcal; range to nearest 10Energy 1.2
Protein per MealBody weight × 1.2–1.6 or 1.4–2.0; ÷ mealskg, gDaily whole g; meals 0.1 gProtein 1.1
Meal contribution comparisonSeparate calorie/protein sums; meal value ÷ complete column total × 100; tied maximumkcal, g, %Shares 0.1%; unknown columns withheldComparison 1.0
Workout Fuel TimingDuration/intensity decision tree; contextual g/kg rangeskg, minutes, hours0.1 g where usefulFuel timing 1.1
Nutrition Label DecoderServing multiplier or portion ÷ 100; amount ÷ DV/RIservings, kcal, kJ, g, mgAmounts 0.1; % wholeLabels 1.2
Added Sugar & DrinksΣ per-serving value × servings; sugar × 4 kcal/gg, kcal0.1 g; whole kcalDrinks 1.1
Caffeine Dose & TimingΣ mg; daily total ÷ kg; circular time-to-bed; disclosed caution rulesmg, kg, timeWhole mg; 0.01 mg/kgCaffeine 1.1
7-Day Sleep ShortfallRecorded-night shortfall and average; unknown nights excluded; complete-log three-night trendhours0.01 hourSleep 1.1
Recipe Yield / GroceryIngredient sum × portion ÷ measured finished mass; ceil(missing grams ÷ edible pack grams) × entered priceg, kcal, currencyNutrients/amounts 0.1; cost 0.01; unknown totals withheldRecipe Ledger 1.0
Grocery & Meal PrepPer-serving × servings × buffer − pantry; round to packuser-selected units0.01 unitGrocery 1.1

US and EU/UK remain separate

US mode uses per-serving arithmetic and current adult/age-4+ Daily Values. EU/UK mode uses per-100 g or ml arithmetic and adult Reference Intakes. US protein is shown in grams without an automatic %DV, EU/UK kJ is entered separately from kcal, added sugar is never inferred from “of which sugars”, and salt is not compared directly with sodium.

The references behind each method

03

Nutrition and Athletic Performance

Academy of Nutrition and Dietetics, Dietitians of Canada and ACSM · 2016 · Used by Protein per meal

Supports: Active-adult protein ranges and meal context
12

Foundation Foods documentation

US Department of Agriculture · Current · Used by Added sugar & drinks tally

Supports: General carbohydrate energy factor context
16

Sleep and heart-healthy living

National Heart, Lung, and Blood Institute · Current · Used by Sleep shortfall planner

Supports: Adult 7–9-hour public guidance
18

Leftovers and food safety

USDA Food Safety and Inspection Service · Current · Used by Grocery & meal prep

Supports: 3–4-day leftovers guidance

Rules we disclose rather than disguise

  • Broad PAL ranges create a starting maintenance interval; they do not measure activity.
  • Fourteen-day calibration and the ±0.25% trend band are planning heuristics.
  • The 300 g/day protein warning is a display heuristic, not a safety threshold.
  • A caffeine dose of at least 100 mg within six hours of bedtime triggers a conservative context flag, not a sleep prediction.
  • A rolling 60-minute caffeine total of at least 1,200 mg triggers an urgent exposure message. The one-hour window is an implementation heuristic for rapid intake, not a regulatory cutoff.
  • A ±0.25-hour sleep trend band labels a week as up, down or roughly steady.
  • Grocery buffer and pack rounding are logistics choices, not nutrition recommendations.

How calculations are checked

Each calculation has a hand-checked reference fixture. Across the suite, tests cover invalid, zero, negative, missing, extreme and unit-conversion cases. Tests reject NaN, Infinity, silent clamping and negative output where it cannot be meaningful. Editorial examples are recalculated when method code changes.

To report an error, include the URL, inputs, expected result, observed result and source if available. See the corrections workflow.

Recent corrections and releases

Recipe Ledger 1.0 and calculator interaction corrections

Added measured recipe yields, basis-matched ingredient estimates, weighed portions, edible-weight pack rounding and reader-entered costs. Unknown nutrition and prices remain incomplete; purchase cost is separate from allocated ingredient cost. Existing forms now clear stale results after input or row edits, kg/lb switching converts the entered weight, unsupported context choices are rejected, and fuel-timeline text follows the actual short-turnaround conditions.

Meal contribution comparison 1.0

Added reader-entered meal calorie and protein totals, separate percentage shares and tied rankings. Blank values remain unknown; incomplete columns have no percentage shares. The Protein 1.1 daily range is unchanged.

Registry corrected; calculator logic unchanged

Synchronized this registry with the live tool pages. Versions now come directly from each tool definition to prevent a separate table falling behind again.

Energy 1.2, Labels 1.2 and Sleep 1.1

Energy and Label tools gained public arithmetic reference fixtures and clearer output reports. Sleep 1.1 distinguishes unknown nights from recorded zero, reports partial-log coverage and does not fill missing nights or extrapolate a full-week total.