-
-
Caregiver Portal authentication screen featuring 1-click Judge Quick Access for instant demo evaluation
-
Dual Frame environment pairing the Echo Show 10 glanceable UI with the real-time Alexa MCP Agent Console.
-
1-Click Mock Voice Dialogue Simulator showcasing 5 hands-free clinical conversation scenarios.
-
Continuous biometric blood glucose tracking dashboard with automated clinical observations.
-
Longitudinal 30-day systolic and diastolic blood pressure trend analysis.
-
30-day eMAR electronic medication administration record and adherence streak matrix.
-
Longitudinal resting heart rate (BPM) telemetry tracking with automated clinical trend observations.
-
High-contrast 6-foot glanceable Bedside Desk Mode with live environment widgets and upcoming dose reminders.
-
Critical cardiac emergency dispatch with automated Ring smart deadbolt unlock for first responders.
-
Hospital-grade A4 Clinical Summary Sheet preview featuring 30-day vitals trajectories and FHIR R4 interoperability.
-
Adaptive responsive layout simulating mobile view for on-the-go family caregiver management.
-
Simulated 850nm IR night-vision Ring Doorbell camera with Computer Vision parcel bounding box telemetry.
-
Monetization tier & paywall modal providing unlimited prescription tracking and certified doctor reports.
-
Real-time dual-turn voice dialogue with active speech bubbles and responsive bottom audio LED glow.
💡 Inspiration: From a Summer in Ea Kar to a Distant IT Job — The Evolution of CareBridge
A while ago, when I was still a student at UIT (University of Information Technology) in Ho Chi Minh City, an astrology website I had spent six months building collapsed overnight after a Google search update. Burned out, broke, and questioning my path, I packed a backpack and took the bus back to my hometown in Ea Kar, Dak Lak (a rural farming district in Vietnam) to stay with my grandmother.
Living with her for several months, I saw a quiet, everyday struggle that never crossed my mind in the city.
My grandma lives with diabetes and severe high blood pressure, taking 6 to 8 different pills every day. Her eyesight is fading with age, and her fingers tremble with arthritis. Every morning, I watched her hold blister packs up to the sunlight, squinting to read tiny labels, terrified she might double-dose or take the wrong pill. Whenever a tiny tablet slipped from her shaking fingers onto the floor, she would apologize to me—as if growing old were her fault. I built the first version of CareBridge back then as a mobile app to help her manage her medicine, and for a while, while I was living under the same roof, it brought her comfort.
Fast forward to today: I graduated from UIT and am now working full-time as an IT software engineer in the city.
Living and working hundreds of miles away from Ea Kar, a completely new reality set in. At the office, staring at lines of code, my mind constantly drifts back home: Did grandma remember her morning blood pressure pill? Did she feel dizzy today? Why hasn't she picked up my call?
That was when I realized the hard truth about mobile apps for elderly care: A smartphone only works if the senior keeps it with them.
Back home in the countryside, my grandma doesn't carry a smartphone in her pocket like young people do. She often leaves it in a jacket pocket in another room, forgets to charge it for days, or doesn't hear the ringtone. Whenever the phone rings, she has to shuffle across the house to find it, squinting at the screen with stiff fingers.
Sitting at my desk as an IT engineer, I asked myself: Why am I expecting an 78-year-old grandmother to adapt to modern personal computers and mobile phones, when technology should adapt to her environment?
She didn't need another app to tap. She needed Ambient Computing—an always-on, always-listening station plugged into the wall 24/7 on her bedside table or living room counter. An ambient display like the Amazon Echo Show 10, where she doesn't have to search for a phone or wear glasses. She just wakes up, glances across the room at oversized numbers, or simply speaks: "Alexa, what pills do I take today?"
And for an IT engineer working far away like me, it provides ultimate peace of mind: knowing that if stock runs low, Amazon Pharmacy can restock it; if she feels chest pain, AWS Bedrock and SNS will immediately dispatch an alert to my phone; and through Ring, I know she is safe without having to call and nag her every day.
That is why I built CareBridge Ambient OS for the Amazon Developer Hackathon. It is the continuation of my promise to my grandmother—bridging the miles between an IT office in the city and a quiet house in Ea Kar.
🌟 What It Does
CareBridge Ambient OS turns bedside smart displays into a 24/7 proactive healthcare companion:
- 6-Foot Glanceable Bedside Display (10-Foot UI): High-contrast oversized clock, punch-card regimen indicators, and circadian sleep-friendly dark teal mode designed for aging eyes across the bedroom.
- Multimodal Voice-First Copilot (Alexa+ MCP): Hands-free vocal check-in. Alexa reads the schedule, confirms intake, and offers a massive 56px+ tremor-tolerant touch button (
"I TOOK MY PILL") with celebration audio feedback. - Web Audio Reactive Cyan Light Bar (
#00CAFF): Signature Echo Show light bar that undulates dynamically along the bottom bezel based on real-time vocal frequency and AWS Polly neural speech. - Autonomous Amazon Pharmacy Refills: When medication inventory drops to $\le 5$ tablets, Alexa detects the shortage and initiates a 1-click refill via Amazon Pharmacy Prime 2-Day delivery.
- Ring Doorbell Smart Cam & Paramedic Rescue Deadbolt: Connects to Ring Doorbell Pro with simulated 850nm IR night-vision, sweeping radar, and computer vision tracking the pharmacy parcel on the porch. In acute emergencies, Ring Smart Deadbolt automatically unlocks for incoming paramedics.
- Geriatric Beers Criteria Safety Shield: Real-time 15-drug pharmacology registry checking 20 dangerous drug-drug interactions (e.g., Warfarin + Aspirin bleeding hazards) before doses are logged.
- One-Tap Doctor A4 Clinical Summary: Instantly renders a printable A4 clinical summary with a 30-day blood pressure trend curve and a scannable HL7/FHIR QR code for hospital visits.
🏗️ How We Built It
CareBridge is architected as an enterprise-grade full-stack TypeScript monorepo:
- Alexa+ Tri-Pillar MCP Server: Built with Express and the official Model Context Protocol SDK (
@modelcontextprotocol/sdk). Implements Streamable HTTP (SSE) (/sse,/message), exposing 7 MCP Tools (getTodaySchedule,logDoseStatus,orderRefill,clinicalAdvisor,recordVitals,ringDeviceHub,negotiateAdherence), 2 MCP Resources (carebridge://...), and 2 Structured MCP Prompts. - Clinical AWS Bedrock Pipeline:
- Claude Haiku 4.5 Native Tool-Use on AWS Bedrock for structured reasoning.
- Bedrock Clinical Guardrails: Topic Denial blocking unapproved cardiac dose alterations and automated PII Redaction.
- Low-Latency Streaming Token Inference: Chunks streamed with
InvokeModelWithResponseStreamCommandcoupled with AWS Polly Neural Voice (Ruth) for a Time to First Audio (TTFA) < 400ms. - AWS SNS Emergency Dispatch: High-priority transactional SMS alerts delivered directly to remote family.
- Hardware-Fidelity Ambient Frontend: Built with Next.js 15 (App Router), React 19, and Tailwind CSS. Calibrated against official Amazon Devices Builder Tools (ADBT) Echo Show 10 display boundaries.
- Atomic SQLite WAL Storage: Local persistence powered by
better-sqlite3in Write-Ahead Logging (WAL) mode for resilient offline execution.
🧗 Challenges We Ran Into
- Bedrock Cross-Region Inference Profiles: In regions like Sydney (
ap-southeast-2), direct model ARN calls were rejected. We built pre-flight profile mapping to route inference through valid cross-region profile IDs (au.anthropic.claude-haiku-4-5-20251001-v1:0). - Audio Feedback Echo Loops on Smart Displays: Echo Show microphones picked up Alexa's own synthesized speech. We engineered an audio mutex (
isBusyRef) with microphone mute locks tied strictly tospeechSynthesis.onendevents. - Streamable HTTP (SSE) Multi-Turn Continuity: Handling real-time SSE stream reconnections without dropping context required building an explicit server-side memory transport registry (
Map<sessionId, SSEServerTransport>) backed by optimistic REST fallback endpoints. - Geriatric Medication Refusal (“Sarah Circuit-Breaker”): When seniors refuse medicine due to nausea or mood, our multi-turn negotiation engine (
negotiateAdherence) steps in. If refusal persists past 2 turns, it triggers a gentle safety circuit-breaker alert to remote family via AWS SNS without confrontation.
🏆 Accomplishments That We're Proud Of
- 100% Passing Automated Test Suite: 73 automated Vitest tests passing across 7 test files, verifying everything from Bedrock Guardrails to multi-tenant SQLite WAL isolation.
- Zero-Downtime Offline Fallback: When cloud connection drops, CareBridge gracefully falls back to deterministic heuristic triage—never freezing or crashing.
- Bridging the Amazon Ecosystem: Unifying Echo Show 10, Alexa+ MCP, Ring IoT, Amazon Pharmacy, and AWS into a cohesive, life-saving ambient experience.
- Detailed 10-Entry Friction Log: Documented authentic developer friction entries to help Amazon and AWS teams refine their next-generation developer tooling.
📚 What We Learned
- Ambient Computing is Compassion in Code: True accessibility for the elderly isn't about bigger touch buttons; it's about eliminating the need to search for a device. An ambient display plugged into the nightstand meets seniors right where they live.
- Clinical AI Demands Strict Guardrails: In healthcare, an unguided LLM can be fatal. Guardrails, deterministic drug-interaction engines, and clear human-caregiver escalation paths are absolute necessities.
- Technology Bridges Distance: As an IT engineer working far from my hometown, building software for the person I love gave every commit, test, and line of code true meaning.
🚀 What's Next for CareBridge Ambient OS
- Physical Echo Show 10 Hardware Deployment: Expanding from our pixel-perfect hardware simulator to live sideloaded production testing on physical Echo Show 10 rotating units.
- Ring Radar Fall Detection: Integrating Ring 3D Motion Detection to detect sudden falls on bedroom floors without requiring cameras or wearable pendants.
- Hospital EHR Integration: Connecting our HL7/FHIR QR engine directly to rural clinic EHR systems so doctors can instantly import compliance histories during checkups.
Built With
- alexa
- amazon-echo-show
- amazon-pharmacy
- aws-bedrock
- aws-polly
- aws-sns
- claude-haiku
- express.js
- healthcare
- kiro-crew
- mcp
- model-context-protocol
- nextjs
- node.js
- react
- ring
- sqlite
- tailwindcss
- typescript
- vitest
- web-audio-api
Log in or sign up for Devpost to join the conversation.