ElderMuscle

Inspiration

Muscle loss kills quietly. 1 in 3 people over 65 suffer from sarcopenia — yet most never get diagnosed until it's too late. We watched elderly family members shrink in strength year by year, eating the same meals, unaware their protein intake was dangerously low for their age. Every health app on the market uses generic adult benchmarks. None of them speak to a 72-year-old's body. We built ElderMuscle to change that.

## What We Learned

Elderly nutrition is clinically distinct. A healthy adult needs 0.8g of protein per kg of body weight — but elderly adults need up to 2.0g depending on their muscle index (SMI). We learned to read InBody data through a geriatric lens, and discovered that when protein is consumed matters as much as how much — spreading intake across meals is critical for muscle synthesis in aging bodies.

## How We Built It

ElderMuscle is a Next.js web app powered by an AI agent backend. Users input their InBody scan data once — the agent computes their SMI and sets an elderly-specific daily protein target. Each meal, users snap a photo. A vision AI identifies the food, estimates protein content, and updates their daily tracker. At week's end, the agent auto-generates a caregiver summary report. Built with Vercel AI SDK, Supabase, and deployed on Vercel.

## Challenges

Translating clinical geriatric nutrition guidelines into reliable AI logic was harder than expected. Food photo recognition varies wildly across cuisines — a Korean doenjang jjigae and a French cassoulet both have hidden protein sources the model had to learn to catch. We also had to design an interface simple enough for elderly users with limited tech experience, while keeping the agent logic sophisticated underneath. Accomplishments That We're Proud of

We built a clinically grounded tool — not just another generic health app. ElderMuscle is the first AI agent that interprets InBody data through a geriatric lens, setting protein targets based on actual SMI scores rather than body weight alone. We're proud that a 70-year-old with no tech background can use it in under 60 seconds: scan, snap, done. The automatic caregiver report closes a gap that no existing app addresses — keeping families informed without burdening the elderly user.

What We Learned

Elderly nutrition is clinically distinct from adult nutrition. We learned that protein timing matters as much as quantity — spreading intake across meals is critical for muscle synthesis in aging bodies. We also learned that designing for elderly UX requires stripping everything down to its absolute core. Every extra button is a barrier. Simplicity is not a design choice here — it's a medical necessity.

## What's Next for ElderMuscle

Wearable integration — pull activity data from smartwatches to adjust protein targets dynamically based on daily movement
Doctor dashboard — give clinicians a longitudinal view of patient muscle health between hospital visits
InBody API partnership — direct data sync from InBody devices, eliminating manual input entirely
Multilingual support — deploy across Japan, Germany, and Italy, the world's most rapidly aging populations
Clinical trial — partner with a geriatric care center to validate ElderMuscle's impact on sarcopenia outcomes at scale