• 

• 

• 

• 

• 

• 

• 

• 

FirstFive

The first five minutes save lives.

An AR + AI voice first-response trainer that turns every student into a capable responder.

---

🏆 CSC Back-to-School Hackathon

Built for real school-life impact — Education · Social Good · Beginner Friendly

---

🚨 In five minutes, a hallway can become a memory.

Ask most students what needs fixing in their school and they'll say the bell schedule 🔔, three tests on one day, or the lunch line. Ask a student who watched someone collapse — while everyone stood frozen 😶 — and the answer changes.

📚 Schools teach algebra, history, and literature. They don't teach students what to do when someone stops breathing. Every school has a cafeteria, a hallway, and a field. Every one of them is a place where a medical emergency can happen. And the person standing closest is almost never a trained professional It's a student. With about five minutes on the clock.

🚨 The Gap Nobody Teaches

First-response training today is a slideshow, a poster, or a CPR mannequin in a closet 🗄️. It's passive, forgettable, and impossible to practice on demand. So when the moment comes, a panicking student has nothing to follow — no guide telling them what to do right now, on this person, in this room.

That gap is a school problem. It just doesn't show up on a schedule.

🎯 What FirstFive Actually Does

FirstFive turns the phone already in a student's pocket 📱 into a spatial, voice-guided first-response coach. Point the camera at a person and the app watches their body in real time, builds a live 3D skeleton, and anchors life-saving instructions onto the exact joints and locations that matter — where a hand goes for compressions, where an EpiPen goes on the thigh, where a tourniquet belongs, where to check a face for drooping.

This isn't a quiz or a flashcard deck 🃏. It's instruction delivered onto a real human body, while both hands are busy and eyes stay off the screen.

🫀 Four Protocols. One Phone. Zero Excuses.

Protocol	What it does
🫀 Cardiac Arrest	A six-step CPR workflow with a pulsing chest marker anchored to the spine joint — so it physically follows the person — plus a live compression counter locked to 100–120 per minute.
😮💨 Allergic Reaction	EpiPen guidance with a pulsing X marker on the thigh and a safe-position baseline that flips from red to green once the person is correctly positioned.
🧠 Stroke	FAST assessment — a live face mesh reads smile blend shapes for symmetry, and orange/blue arm targets reveal as each arm is raised to test for weakness.
🩸 Injury & Bleeding	Tourniquet trainer — tap to mark the wound, get a safe-zone line 2–3 inches above it, follow rotating arrows for wrap direction, then track elapsed tourniquet time.

Every protocol is a guided sequence, not a wall of text — telling the responder the next step the moment they need it.

🏗️ Under the Hood

┌──────────────────────────────────────────────────────────────┐
│                       STUDENT'S iPHONE                        │
│  ┌─────────────────────┐       ┌──────────────────────────┐  │
│  │   SwiftUI UI LAYER   │◄─────►│  EMERGENCY / TRAINING     │  │
│  │  Home · Protocols    │       │  mode router              │  │
│  │  Vitals panel        │       └──────────────────────────┘  │
│  └──────────┬───────────┘                                     │
│  ┌──────────▼──────────────────────────────────────────────┐ │
│  │            ARKit BODY TRACKING (rear camera)             │ │
│  │   ARBodyTrackingConfiguration  →  ARBodyAnchor           │ │
│  └──────────┬──────────────────────────────────────────────┘ │
│  ┌──────────▼──────────────────────────────────────────────┐ │
│  │              RealityKit SPATIAL ENGINE                    │ │
│  │  • 19-segment skeleton (Bones enum)                       │ │
│  │  • Heatmap joint spheres + oriented bone meshes           │ │
│  │  • Posture detection: spine vector → "is the person lying  │ │
│  │    down?" → unlocks safe-position feedback                 │ │
│  │  • Joint-anchored markers: chest · EpiPen · tourniquet     │ │
│  └──────────┬──────────────────────────────────────────────┘ │
│  ┌──────────▼──────────────────────────────────────────────┐ │
│  │              AI VOICE GUIDANCE LAYER                      │ │
│  │  scripted segments → emotion profile → ElevenLabs TTS     │ │
│  │  → AVAudioPlayer sequential playback + waveform UI        │ │
│  └──────────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────┘

Built with: Swift 5 · SwiftUI · ARKit (body + face tracking) · RealityKit · ElevenLabs emotion-aware text-to-speech · AVFoundation · Xcode, targeting iOS 18+ on ARKit-capable devices.

AI-use disclosure: AI coding and design assistants (ChatGPT, Claude, Gemini) were used for brainstorming, scaffolding, debugging, and design exploration. The architecture, protocol content, AR approach, and final product were designed, assembled, and are fully understood by the team, who can explain every component and decision.

🎓 Practice. Then Perform.

FirstFive ships with two modes because knowing and doing are different skills.

🚑 Emergency Mode is direct and urgent — calm, hands-free narration counts a student through compressions in real time, with a live waveform showing when the voice is speaking.

🎓 Training Mode is the rehearsal: the same protocols and the same spatial overlays, with completion tracked per scenario, so a student can master all four before they ever need one. Practice when it's quiet. Perform when it isn't.

📊 What Changes When Every Student Is Ready

Not a feature. A room that knows what to do.

😰 Before FirstFive	😌 After FirstFive
📊 First response taught as a slideshow	🕶️ Live AR guidance on a real body
📌 A poster nobody reads	🦴 A 3D skeleton tracking the person in front of you
🗄️ A mannequin in a closet	📱 A coach in every pocket
🥶 A crowd that freezes	🗣️ A calm voice saying the next step

Before, five minutes pass in panic. After, with a plan. 🧭

🌍 Why This Matters Beyond One School

Not a niche tool — the same fix everywhere:

• 🏫 Every school has the same gap. Training is under-taught.
• 🧑🤝🧑 Students are always the closest responders. Not the nurse — the kid right there.
• 📱 No hardware needed. No mannequins, no kits, no budget. Just a phone students already carry.
• 🌐 It scales to any school, country, or language. Universal protocols.
• ⏱️ The five-minute window is the same everywhere. The clock doesn't care where you live.

A bell-schedule app fixes one school. This fixes the moment in all of them!

🛠️ What Was Hard to Build

Not a thin wrapper. Every layer is engineered:

• 🦴 Raw ARKit joint data into a readable skeleton. 19 bone segments, heatmap joints, 3D meshes.
• 🎯 Markers that stay stable as the person moves. Chest, EpiPen, and tourniquet markers anchor to joints, following a breathing body instead of drifting.
• 🗣️ An emotion-aware voice that shifts tone. Calm while teaching, urgent when seconds count.
• 🫀 Four protocols, each with its own spatial logic. A CPR rhythm counter, a face-mesh symmetry check, a thigh injection site, a rotating wrap.
• 🧭 Posture detection from geometry. A spine vector decides whether the person is lying down — unlocking the next step only when it's safe.

Real sensors. Real geometry. Real time.

⏱️ The first five minutes save lives.

Schools already ask students to be responsible for each other. FirstFive just gives them something to actually follow. The closest person is almost never a professional — so make the closest person ready.

Built With

• 3d-skeleton
• accessibility
• arkit
• augmented-reality
• avfoundation
• body-tracking
• computer-vision
• education
• elevenlabs
• emergency-preparedness
• face-tracking
• first-response
• hackathon
• health-tech
• ios
• realitykit
• social-good
• swift
• swiftui
• text-to-speech
• xcode