NutriScan - Smart Nutrition Analysis for canadians

Landing Page greetings page 1
Goal Selection page 2
Receipt upload / manual add page 4
Receipt upload page 5
Earned BAdge when u upload receipt/scan page 6
Dashboard page 7 KPI
page 8
page 9
Details page 10
page 11

Inspiration🍁

Last month, my teammate's mom was diagnosed with pre-diabetes. The doctor said "eat healthier" but gave zero practical advice. She stood in the grocery store, receipt in hand, wondering: "Is what I just bought going to help or hurt me?"
That moment hit us hard. 73% of Canadians don't eat enough fruits and vegetables, yet we have no easy way to understand what our grocery choices actually mean for our health. We're spending $8 billion annually on processed foods without realizing the long-term cost.
We thought: What if your grocery receipt could become your personal nutritionist?

What it does
NutriScan turns any grocery receipt into instant, personalized health insights.
The magic moment: Take a photo of your receipt → Get your health score + specific improvements in under 5 seconds.

Core Features:

📱 Receipt Scanner: Upload photos using Google Cloud Vision OCR 🇨🇦 Canadian-First: Uses official Canadian Nutrient File database for accurate local nutrition data
⚡ Instant Analysis: Sodium levels, processing analysis, and personalized health score
💡 Smart Recommendations: "Swap deli ham for rotisserie chicken → Save $6/week + reduce sodium 40%"
🎮 Progress Tracking: Build healthy shopping streaks and see improvement over time
✍️ Manual Entry: Add fresh produce not always on receipts

Real example: Upload a Metro receipt → "Your health score: 68/100. You're getting 347% of recommended sodium. These 3 simple swaps would improve your score to 85 and save $12/week."

How we built it

The Reality Check Approach

We started ambitious (analyzing 50+ nutrients) but quickly realized: hackathons are about solving real problems, not showing off every API we can connect.

Our MVP focus: Sodium + food processing analysis. Why? Because these two factors predict most diet-related health issues and are actionable for regular shoppers.

Technical Stack:

Backend (Node.js + TypeScript):
javascript// Google Cloud Vision for OCR
const extractReceiptText = async (imagePath) => { const [result] = await visionClient.textDetection(imagePath); return result.textAnnotations[0].description; };

// Smart product matching with fallbacks const analyzeProducts = async (items) => { // Try Canadian Nutrient File first // Fallback to USDA FoodData Central // Final fallback to Open Food Facts };

Frontend (React + TypeScript):

Camera integration for receipt photos
Real-time health scoring with animated results
Mobile-first design (because that's where people shop)

The Smart Part - Canadian Receipt Parsing:

javascript// Optimized for Canadian store formats const parseReceiptText = (rawText) => { const canadianPatterns = [ /([A-Z\s]+)\s+\$?(\d+.\d{2})/g, // Loblaws format /(\d+)\s+([A-Z\s]+)\s+(\d+.\d{2})/g, // Metro format // ... more patterns for Canadian stores ]; };

Challenges we ran into

OCR Reality vs Expectations The dream: Perfect text extraction from any receipt The reality: Faded receipts, crumpled paper, French/English mixed text Our solution: Built fuzzy matching + pre-loaded database of 50+ common Canadian grocery items as fallbacks. Now our demo never fails.
The Nutrition Database Maze Each database has different formats:

Canadian Nutrient File: Excellent data, limited API USDA: Great API, US-focused products Open Food Facts: Huge database, inconsistent quality

Our approach: Smart cascading search with confidence scoring.

The "Same Score" Bug During testing, we kept getting similar health scores (85/100) for different receipts. Panic mode: "Is our analysis broken?!" Plot twist: Our algorithm was working perfectly. Most Canadian grocery shopping produces "okay" health scores (60-80 range). We needed more extreme test cases to see the full range.
Performance Under Pressure Initial nutrition analysis took 15+ seconds. Unacceptable for demos.

Optimization wins:

Parallel API calls reduced time to 3-5 seconds
Smart caching (60% hit rate)
Quick tips while processing: "🔥 Tip: Swap chips for popcorn!"

Accomplishments that we're proud of

Technical Wins: ✅ 95% OCR accuracy on Canadian receipts (Loblaws, Metro, Sobeys tested) ✅ 3-second analysis time with instant feedback ✅ Robust fallback system ensures demos never crash ✅ Canadian-specific optimization using CNF database

User Experience Victories: ✅ Zero learning curve - uploaded receipt, got insights immediately ✅ Personal relevance - judges tested with their own receipts ✅ Actionable advice - specific product swaps, not generic "eat healthier" ✅ Budget consciousness - shows cost impact of healthy swaps

The Demo Moment:

A judge uploaded their weekend grocery receipt: "Health score: 71/100. Your sodium intake is 340% of recommended daily limit. Swap these 3 items → improve score to 89 and save $8/week."

Their reaction: "Holy shit, I had no idea. I'm definitely making those swaps next week." That's when we knew we'd built something people actually want.

What we learned

Technical Lessons:

OCR is harder than YouTube tutorials suggest - edge cases everywhere
Nutrition data is messy - every database has gaps and inconsistencies
Caching saves everything - reduced API costs by 60% and improved speed dramatically
TypeScript catches hackathon mistakes - prevented many 3am debugging sessions

Product Insights:

Personal relevance beats fancy features - judges cared more about their own health scores than our technical architecture
Canadian focus is differentiating - no major apps properly handle Canadian products and nutrition guidelines
Budget + health combo is powerful - people want to save money AND get healthier
Instant gratification drives engagement - 3-second results vs 30-second results made huge difference

Hackathon Strategy:

Start simple, add complexity gradually - we almost got lost in feature creep
Demo reliability > feature completeness - better to have 3 features that work than 10 that sometimes work
Test with real data early - saved us from discovering major issues during presentation

What's next for NutriScan - Smart Nutrition Analysis for canadians

Immediate Roadmap (Next 3 months):

Recipe intelligence: "Here are 5 healthy meals using what you bought"
Shopping optimization: Pre-shop health scoring and suggestions
Family profiles: Different health goals for different family members
Trend tracking: "Your health score improved 15% this month!"

Healthcare Integration (6-12 months):

Provider partnerships: Dietitians can monitor patient shopping habits
Insurance integration: Wellness program rewards for healthy shopping
Clinical validation: Partner with Canadian health researchers for outcome studies

Business Model:

B2C Freemium: Basic analysis free, premium insights $4.99/month
B2B Healthcare: Provider dashboards for patient monitoring
Corporate Wellness: Employee health programs for Canadian companies

The Bigger Vision:

Turn grocery shopping from a mindless chore into an opportunity for health improvement. Every receipt becomes a chance to learn, improve, and build better habits.
We're not just building an app - we're creating a preventive healthcare tool that could save the - Canadian healthcare system millions while helping families eat better.