惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Apple Machine Learning Research
Apple Machine Learning Research
J
Java Code Geeks
博客园 - 聂微东
Microsoft Azure Blog
Microsoft Azure Blog
量子位
T
Tailwind CSS Blog
Vercel News
Vercel News
I
InfoQ
Stack Overflow Blog
Stack Overflow Blog
U
Unit 42
Engineering at Meta
Engineering at Meta
L
LangChain Blog
大猫的无限游戏
大猫的无限游戏
D
Docker
博客园_首页
P
Proofpoint News Feed
月光博客
月光博客
T
The Blog of Author Tim Ferriss
MyScale Blog
MyScale Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
Martin Fowler
Martin Fowler
腾讯CDC
N
Netflix TechBlog - Medium
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
How We Built an Adult Height Predictor: The Math, the Sci...
Fitness Tools · 2026-06-15 · via DEV Community

Fitness Tools

Ever wonder how those "how tall will my child be?" calculators actually work? They're not magic — but the science behind them is surprisingly interesting. In this article I'll break down the methodology behind adult height prediction, walk through how to implement it cleanly on the web, and share the lessons we learned building the Adult Height Predictor at Nutryio.
The Science: Mid-Parental Height Method
The most well-validated approach to predicting adult height is the mid-parental height (MPH) formula, originally described by Tanner et al. in growth studies from the 1970s and refined repeatedly since. It's still the standard used in clinical pediatric practice today.
The formula is beautifully simple:
// For males:
MPH = (father_height + mother_height + 13) / 2

// For females:
MPH = (father_height + mother_height - 13) / 2

The 13 cm offset accounts for the average height difference between adult males and females. The result is the target height — the genetic "midpoint" a child is likely to grow toward.
The Prediction Range
No single number tells the whole story. Clinical guidelines add a ±8.5 cm corridor around the mid-parental target, giving a more honest low-to-high prediction range.

This range captures roughly 95% of children from a given parental height combination.
Adding Current Age: Growth Remaining
The MPH tells you the destination. But users also want to know: how much growing is left?
This requires factoring in the child's current age and sex. Growth plates (epiphyseal plates) close at different times:
Sex
Peak Growth Spurt
Final Height Age
Boys
12–15 years
17–19 years
Girls
10–13 years
15–17 years
A rough model for growth remaining as a function of age:

In practice, growth isn't perfectly linear — there's a sigmoid-like acceleration during puberty — but for a web calculator aimed at parents rather than endocrinologists, this gives a usable approximation.
Putting It Together: Full Implementation
Here's a clean, self-contained implementation in vanilla JS:
/**

  • Adult Height Predictor
  • Based on mid-parental height method (Tanner et al.) */ function predictAdultHeight({ sex, ageYears, currentHeightCm, fatherHeightCm, motherHeightCm }) { // Step 1: Mid-parental height target const offset = sex === 'male' ? 13 : -13; const mph = (fatherHeightCm + motherHeightCm + offset) / 2;

// Step 2: Prediction range (±8.5 cm, ~95th percentile corridor)
const low = mph - 8.5;
const high = mph + 8.5;

// Step 3: Estimated growth remaining
const maturityAge = sex === 'male' ? 18 : 16;
const growthLeft = Math.max(0, mph - currentHeightCm);
const yearsRemaining = Math.max(0, maturityAge - ageYears);

return {
midParentalTarget: Math.round(mph * 10) / 10,
lowPrediction: Math.round(low * 10) / 10,
highPrediction: Math.round(high * 10) / 10,
growthRemainingCm: Math.round(growthLeft * 10) / 10,
yearsUntilMaturity: yearsRemaining,
};
}

UX Lessons from Building This

  1. Always show a range, never just a point estimate Showing a single number like "your child will be 178 cm" sets false expectations. Growth is inherently probabilistic. Displaying a low/high range builds trust and reflects the actual science.
  2. Localise units — but store in metric Users in the US think in feet and inches; the rest of the world uses centimeters. We store everything internally in cm and convert for display:
  3. Frame the result, not just the number "Your child may reach between 170–187 cm" is just data. Add context: where does this fall on population growth charts? Is the current height tracking ahead, behind, or in line with the prediction? That's what parents actually care about.
  4. Be upfront about limitations Height prediction calculators typically have an accuracy window of ±5–7 cm in real-world populations. Hormonal conditions, chronic illness, nutrition, and sleep all affect actual outcomes. A responsible tool states this clearly — near the result, not buried in an FAQ. What the Calculator Can't Know This is worth its own section. The mid-parental method is well-validated on average across populations, but there are several factors it cannot account for: Growth hormone deficiency or excess — endocrine conditions can significantly alter growth trajectory Chronic illness — conditions affecting nutrition absorption (e.g. coeliac disease) suppress growth Delayed vs. early puberty — a late bloomer might track short at 14 and land perfectly in range at 20 Nutrition and sleep — especially in the 0–5 year window, these have outsized influence on growth potential For children whose growth is tracking far outside their mid-parental range, a paediatrician or paediatric endocrinologist is the right next step — not a recalculation. Try It You can test the live calculator at https://nutryio.fit/calculators/adult-height-predictor. It also links to related tools for growth percentile and pediatric BMI if you want to explore the full picture. Wrapping Up Adult height prediction is a great case study in building honest health tools on the web: there's real science to implement, meaningful UX decisions to make, and important caveats to communicate clearly. The mid-parental height method is accessible enough to implement in an afternoon, yet backed by decades of clinical research. If you're building something similar, the key principles are: Use the MPH formula with a ±8.5 cm range Factor in age and sex for growth-remaining estimates Show ranges, not point predictions Be explicit about what the tool can't account for Questions or corrections? Drop them in the comments — especially if you've built growth-tracking tools yourself. Built something similar? I'd love to see it. Tag me or share a link below.