Browser / CDN

Load two scripts. No build step required. Tables are fetched once and cached by the browser.

Copy<script src="https://flame-cai.github.io/anthro/anthro.bundle.js"></script>
<script src="https://flame-cai.github.io/anthro/api.js"></script>
<script>
  await AnthroAPI.ready()

  const r = AnthroAPI.compute({
    mode:      'day',
    sex:       'female',
    dob:       '2024-01-15',
    weight_kg: 7.0,
    height_cm: 64.0,
    muac_mm:   136,
  })

  r.lhfa            // 'Moderately stunted'
  r.z_lhfa          // -2.7901
  r.wfa             // 'Normal'
  r.muac_threshold  // 'Normal'
  r.bmi_val         // 17.0898
</script>

npm

Copynpm install @flame-cai/anthro

Copyconst { createAnthro } = require('@flame-cai/anthro')
const fs    = require('fs')
const names = ['wfa', 'lhfa', 'bmi', 'acfa', 'wfl', 'wfh']
const load  = (p, n) => JSON.parse(fs.readFileSync(
  require.resolve(`@flame-cai/anthro/data/${p}_${n}.json`)))

const anthro = createAnthro(
  Object.fromEntries(names.map(n => [n, load('day',   n)])),
  Object.fromEntries(names.map(n => [n, load('month', n)]))
)

const result = anthro.compute({
  mode: 'day', sex: 2, dob: '2024-01-15',
  weight_kg: 7, height_cm: 64, muac_mm: 136,
})

Python / PyPI

Copypip install anthro

Copyfrom anthro import compute, batch

result = compute({
    'sex':       'female',
    'dob':       '2024-01-15',
    'measured':  '2025-01-15',
    'weight_kg': 7.0,
    'height_cm': 64.0,
    'muac_mm':   136,
})

result['lhfa']            # 'Moderately stunted'
result['z_lhfa']          # -2.7901
result['wfa']             # 'Normal'
result['muac_threshold']  # 'Normal'

Full Python API reference: python/README.md

Clone directly

Use src/anthro.js and data/ directly without npm. Works in Node.js and any bundler (webpack, Vite, Metro).

Copygit clone https://github.com/flame-cai/anthro.git

Copyconst { createAnthro } = require('./anthro/src/anthro.js')
const names  = ['wfa', 'lhfa', 'bmi', 'acfa', 'wfl', 'wfh']
const anthro = createAnthro(
  Object.fromEntries(names.map(n => [n, require(`./anthro/data/day_${n}.json`)])),
  Object.fromEntries(names.map(n => [n, require(`./anthro/data/month_${n}.json`)]))
)

React Native (Android & iOS)

The library has no platform-specific dependencies. Metro handles JSON require() natively.

Via npm

Copyimport { createAnthro } from '@flame-cai/anthro'

const names  = ['wfa', 'lhfa', 'bmi', 'acfa', 'wfl', 'wfh']
const anthro = createAnthro(
  Object.fromEntries(names.map(n => [n, require(`@flame-cai/anthro/data/day_${n}.json`)])),
  Object.fromEntries(names.map(n => [n, require(`@flame-cai/anthro/data/month_${n}.json`)]))
)

export function assess(child) {
  return anthro.compute({
    mode: 'day', sex: child.sex,
    dob:       child.dateOfBirth,
    weight_kg: child.weightKg,
    height_cm: child.heightCm,
    muac_mm:   child.muacMm,
  })
}

By copying files

Copy src/anthro.js and the entire data/ folder into your project.

Create the instance once at module level, not inside a component render function. The constructor builds O(1) lookup maps (~50 ms on a mid-range device).

Missing inputs

When a measurement is absent the affected indicator returns a descriptive string. All other indicators are still computed.

Copyanthro.compute({ sex: 'F', dob: '2024-01-15', height_cm: 64 })

r.lhfa            // 'Moderately stunted'  -- computed normally
r.wfa             // 'Missing weight to compute'
r.muac_threshold  // 'Missing muac to compute'

Input reference

Output reference

Table sources

Day-indexed tables are extracted from the R package anthro v1.1.0 (WorldHealthOrganization/anthro): weianthro, lenanthro, bmianthro, acanthro, wflanthro, wfhanthro -- the same datasets used in WHO igrowup SAS/SPSS/Stata packages. Month-indexed tables are the WHO supplementary monthly tables from the same study.

Formula

Source: WHO Technical Report 2006, section 5.2, pp. 300-304.

z = [(X/M)^L - 1] / (L x S)   when L != 0
z = ln(X/M) / S                 when L = 0

SD23 adjustment when |z| > 3:
  z = +3 + (X - SD3)  / (SD3 - SD2)    when z > +3
  z = -3 + (X - SD3') / (SD2' - SD3')  when z < -3
  where SD3 = M(1 + L*S*3)^(1/L),  SD2 = M(1 + L*S*2)^(1/L)

Measure correction

Source: WHO 2006, Chapter 7. Mean difference between recumbent length and standing height = 0.73 cm, applied as 0.7 cm.

Day mode: the loh field in lenanthro per day determines the expected type (L for days 0-730, H for days 731-1826). Month mode: L for months 0-23, H for months 24-60.

Classification cut-points

Source: WHO (2009), ISBN 9789241598163.

Plausibility flags

Source: WHO igrowup documentation (igrowup_standard.sas). flag_* = 1 indicates a z-score outside the biologically plausible range; the observation should be reviewed.

Precision

This library uses the exact floating-point z-score for classification without prior rounding. The R anthro package rounds z to 2 decimal places before classifying: a z-score of -2.003 rounds to -2.00 and is classified as Normal rather than MAM. This library retains full precision, which is relevant for observations near a cut-point boundary.

Citations

Changelog

1.1.1

• Python package (anthro on PyPI) — same WHO tables, formula, and output schema as the JS library
• Shared data/ directory — no duplication between JS and Python packages
• Global VERSION file — single source of truth for npm, PyPI, and JS source
• CI publishes both npm and PyPI on version tag push
• 34 Python tests verified against R anthro v1.1.0

1.1.1

• Day-indexed (WHO igrowup) and month-indexed (WHO supplementary) table modes
• Exact floating-point z-scores -- not pre-rounded before classification
• Automatic +/-0.7 cm measure correction from loh field
• 6 indicators: muac_threshold, acfa, bmi, lhfa, wfa, wflh
• Missing input returns a descriptive string, not null
• Day 1826 (exactly 60 months) inclusive in all tables
• TypeScript declarations
• 33 tests verified against R anthro v1.1.0

Contributing

JS tests: node test/anthro.test.js — no install required.

Python tests: python3 python/test_anthro.py — no install required.

To re-export day-indexed tables from the R anthro package: Rscript scripts/export-tables.R

CONTRIBUTING.md · Issues · github.com/flame-cai/anthro