FastTrack

• 

  Patient intake form

• 

  Staff Dashboard

Emergency rooms are chaotic environments where triage decisions directly impact patient outcomes. We were inspired by the unfortunate reality that manual triage is time-consuming, inconsistent, and prone to human error, especially during high-volume times. We wanted to build a tool that could help ER staff prioritize patients faster and more accurately using AI.

FastTrack is an AI-powered ER triage system with two interfaces. Patients fill out a digital intake form capturing their name, contact info, symptoms, pain level, and medical history. The moment they submit, Claude AI analyzes their data and assigns a ranking. The staff dashboard pulls from the same database and displays patients ranked in order of clinical priority, so ER staff always know who to see next.

We built the patient-facing frontend in HTML/CSS/JavaScript and the backend in Python/JavaScript. Patient data is stored locally in JSON. On each form submission, the backend calls the Anthropic Claude API (Haiku model) with a structured clinical prompt to perform AI triage. We also built a rule-based manual triage fallback that kicks in if the AI is unavailable.

We learned how to integrate LLM APIs into a real application with structured output requirements, how to design prompts that return reliable JSON. We also learned a lot about the real-world complexity of ER triage and the ESI (Emergency Severity Index) system.

Built With

• css
• html
• javascript
• python