ClearPath

Inspiration

Hospitals are terrifying. You're sick, you're scared, and sometimes you don't speak the language. You walk through the doors and someone hands you a clipboard. You don't know which building to go to, how long you'll wait, or what's happening to your loved one down the hall. We built ClearPath because the moment someone walks into a hospital is exactly when technology should be most helpful and right now, it's almost entirely absent.

What it does

ClearPath is a full hospital operating system that guides patients from Check-in to Discharge in any language. Patients open the app and have a real conversation with an AI triage nurse that asks follow-up questions, detects their language automatically, assesses their emotional state, and routes them to the right department across any hospital. They get an animated walking map, a live queue position, and real-time status updates pushed directly to their phone saying "Your test is in progress" or "Transport is on the way." On the staff side, a live dashboard shows every patient across every department in real time. Staff can see patient sentiment ("Patient is scared and traveling alone"), detect patients who are stuck waiting too long, update statuses that instantly notify the patient, and generate a full AI shift handoff summary in one click. A third analytics screen shows patient volume by hour, average wait times by department, urgency breakdown, and every language spoken that day.

Stories

True Story #1: My brother was bitten by a cat in Puerto Rico, and what should’ve been simple turned into four hours of confusion. We didn’t speak Spanish. We didn’t know where to go. No one could tell us wait times or whether rabies treatment was even available. Every stop felt like a dead end. Eventually, we gave up and spent $600 on an emergency flight home—just to feel certain. With ClearPath, we could have navigated everything in our own language, known exactly where to go, and confirmed treatment ahead of time. Instead of panic, we would’ve had clarity.

True Story #2: I went to Yale Hospital with an eye infection and waited 12 hours to be treated. What made it worse was the uncertainty. I kept being told I was “almost” admitted, but nothing changed. No timeline, no visibility. Even after admission, I didn’t know where to go and ended up waiting even longer. At one point, I considered leaving without treatment. With ClearPath, I would’ve seen my place in line, gotten real-time updates, and known exactly where to go. Instead of feeling lost, I would’ve felt in control.

How we built it

Frontend — React with Vite, split across a patient app, staff dashboard, and analytics page, Backend — Node.js with Express, handling 12 REST API endpoints and a WebSocket server for real-time communication, Database — SQLite via better-sqlite3, storing patient records, queue positions, statuses, and timestamps, AI — Claude API powering the diagnostic conversation, language detection, sentiment analysis, triage routing, and shift handoff summaries, Real-time — WebSockets connecting patient phones directly to the staff dashboard so status changes push instantly in both directions, Map — Hand-built SVG floor plan of the actual Yale New Haven Hospital campus with animated routing paths, elevator markers, and floor-by-floor navigation

Challenges we ran into

Getting the diagnostic conversation to feel natural was harder than expected. Claude needed to know when to stop asking questions and route. If it asked too few clarifying questione, the routing would be wrong. If it asked too many, it would be ineffective, take too long, and frustrate patients. Tuning that balance took significant prompt engineering. Managing WebSocket connections across multiple patients and the dashboard simultaneously required careful socket mapping so the right message always reached the right device. We also ran into issues with duplicate Express routes silently overriding each other, which caused stuck detection to stop working until we traced the conflict.

Accomplishments that we're proud of

We're proud that the system works end to end. A patient can type in Arabic, get routed to the right building, walk there using the map, watch their queue position update live, and get notified the moment staff calls them. The full journey works in any language with zero configuration. We're also proud of the sentiment tracking. The fact that a staff member can see "Patient is scared and traveling alone" before they even reach the desk changes how that interaction begins.

What we learned

We learned that the hard part of building for healthcare isn't the technology, it's understanding the human moment you're designing for. Every feature we added had to answer the question: does this make a scared person feel less alone? We also learned how much real infrastructure it takes to make AI feel seamless. Claude handles the understanding layer, but the WebSocket server, the database, the status system, the stuck detection makes a product.

What's next for ClearPath

Electronic medical records integration so patient history informs the triage conversation Real indoor positioning using Bluetooth beacons for turn-by-turn navigation inside the hospital Emotional escalation detection and flagging patients in crisis before they finish checking in A family portal so loved ones get live updates without calling the front desk HIPAA compliance and a pilot partnership with a real hospital system

Built With

• and-analytics-page-vite-?-frontend-build-tool-node.js-+-express-?-backend-rest-api-and-websocket-server-sqlite-+-better-sqlite3-?-patient-database-claude-api-(anthropic)-?-diagnostic-conversation
• better-sqlite3
• claudeapi
• dotenv
• express.js
• language-detection
• node.js
• qrcode.react
• react
• sentiment-analysis
• shift-handoff-summaries
• sqlite
• staff-dashboard
• svg
• triage-routing
• uuid
• vite
• websockets