AnthroZ: Child and Adolescent Anthropometric Z-Score Calculator

Abstract

A free, portable, offline Windows application that calculates length/height-for-age, weight-for-age, and BMI-for-age z-scores (standard deviation scores, SDS) and percentiles for children and adolescents from a CSV file. No installation is required: download and extract AnthroZ_Calculator_v6.0_Windows_App.zip, keep its contents together, and run AnthroZ_Calculator.exe. All calculations run locally, and the results, a companion report, and a data dictionary are saved on the user's computer. The application makes no network connections and has no upload, telemetry, or analytics features, so researchers can calculate z-scores without uploading identifiable patient data to a website. Users remain responsible for handling their data in accordance with ethics approvals, institutional policies, and applicable data-protection laws. Growth references WHO (default), birth to 19 years: WHO Child Growth Standards 2006 below 60 months, using exact age in days; WHO Growth Reference 2007 from 60 months. Weight-for-age from 121 months uses the CPEG/Rodd 2014 extension and is explicitly labelled as non-WHO. CDC, 2 to under 20 years: CDC 2000 Growth Charts, with a choice of the original or the 2022 extended BMI-for-age method. Below 2 years, the CDC options use the WHO standards, as CDC recommends. From 2 years of age, the application also reports BMI weight-status categories (underweight, healthy weight, overweight, obesity) using the CDC/AAP percentile cut-offs applied to the selected growth reference; the output column is named bmi_class_cdc_aap accordingly. With the WHO option, these are CDC/AAP-style categories calculated from WHO reference values; they are not WHO's own weight-status definitions, which use z-score cut-offs. The CDC options additionally report obesity classes 1–3, BMI at the 95th percentile, and BMI as a percentage of that value. New in version 6.0: before giving any results, the application checks the whole file for problems that usually mean a unit or column error (heights in metres, inches, or millimetres; weights in grams or pounds; ages in years or days; height and weight columns swapped; dates the wrong way round), names the likely cause as something to check, and lets the user stop or proceed. A data dictionary is saved with every results file, and saved numbers are rounded to meaningful precision. Calculations are unchanged from version 5.7, apart from one rare CDC case corrected (see the changelog). Outputs identify the reference used for each indicator, and include biological-plausibility flags and row-level audit notes where relevant. Results were compared with WHO's own R code and CDC's cdcanthro R package on more than 80,000 synthetic records, with no mismatches, and with CDC's published percentile tables. Two independent audits found no error in the calculations, and every check was repeated on a second computer. The validation scripts and evidence are included. Large files: there is no fixed limit on the number of rows. A synthetic file of 1.2 million children was processed in about 1 minute, using 3.6 GB of memory; memory use grows by about 3 GB per million rows, so the practical limit is the computer's memory. Files AnthroZ_Calculator_v6.0_Windows_App.zip: the portable Windows application only, for most users (12 MB). AnthroZ_Calculator_v6.0.zip: the complete release (35 MB), with the Windows application, the full Python source code, the independent validation materials, the documents (PDF), and SHA-256 checksums for every file. Begin with START HERE.txt. A companion application, the BPZ Calculator (blood pressure z-scores), is available at https://doi.org/10.5281/zenodo.22950422. Licence: The application is released under the MIT licence. The bundled WHO reference files retain their own GPL-3.0 licence, included alongside those files. Disclaimer: This is independent research software. It is not produced, certified, or endorsed by WHO, CDC, or CPEG, and it is not a medical device. Developed by José Guilherme Behrensdorf Derraik, with assistance from Claude Fable 5, Claude Opus 5, and Claude Opus 5.5 (Anthropic), and ChatGPT 5.6 Sol, ChatGPT 6 Astra, and ChatGPT 6 Sol (OpenAI). These AI systems assisted with code, tests, documentation, and audits under human direction and review. They are acknowledged as tools, not authors; the human developer remains responsible for the release. Suggestions and problem reports are welcome. Contact José Derraik at j.derraik@auckland.ac.nz. When reporting a problem, please include the application version and, if possible, a small de-identified example.