MedAssist Triage

🏥 Inspiration

In emergency departments and urgent care clinics across the globe, triage nurses and physicians face high patient volumes, severe cognitive overload, and extreme time pressure. Delayed recognition of critical conditions—such as atypical acute coronary syndrome, acute stroke within the thrombolytic window, or early sepsis—can result in devastating clinical outcomes. We were inspired to build MedAssist Triage to empower healthcare providers and clinical intake desks with instantaneous, standardized, and evidence-backed triage intelligence that ensures critical patients receive life-saving care without delay.

⚡ What It Does

MedAssist Triage acts as an intelligent clinical co-pilot at the front door of emergency care:

1. Intelligent Intake & Biometrics Tracker: Captures patient symptoms, duration, onset, allergies, medications, and vital signs (HR, BP, SpO2, RR, Temp, Pain Score) with real-time hemodynamic safety alerts for shock, severe hypoxia, or hypertensive crisis.
2. Automated ESI Triage (Levels 1 to 5): Evaluates patient stability and resource needs according to the gold-standard Emergency Severity Index (ESI v4) algorithm.
3. Critical Red Flag Detection: Flags urgent life-threatening presentations (e.g. stroke 4.5h thrombolytic window, anaphylaxis airway compromise, sepsis SIRS criteria) with prominent clinical alerts.
4. Physician Handoff Generator: Automatically synthesizes structured SBAR (Situation, Background, Assessment, Recommendation) and SOAP (Subjective, Objective, Assessment, Plan) medical notes, complete with differential diagnosis probabilities and recommended test orders.
5. Emergency Department (ED) Live Queue Dashboard: Provides a real-time provider board to track waiting patients, bed allocations, and clinical status (Waiting → In Triage → Physician Assigned → Under Treatment → Discharged).
6. Emergency Directives & Hotlines: Includes quick-access emergency protocols for CPR/AED, Stroke FAST, and 1-hour Sepsis bundles.

🛠️ How We Built It

• Frontend & UI: Built with React 19, TypeScript, and Tailwind CSS, styled in a hospital-grade dark glassmorphism aesthetic with glowing vital widgets.
• Triage Decision Engine: Developed an evidence-based clinical reasoning engine implementing ESI v4 decision logic, differential scoring, and SBAR/SOAP note synthesis.
• AI & LLM Inference Layer: Designed a hybrid architecture that runs seamlessly offline via built-in clinical intelligence, with plug-and-play integrations for Featherless AI, Google Gemini, and OpenAI.
• Build & Tooling: Powered by Vite 6 for blazing fast compilation and zero-lag production bundling.

🧗 Challenges We Ran Into

• Balancing rapid intake data entry with the clinical depth needed for high-fidelity triage assessment.
• Implementing accurate vital sign thresholds and hemodynamic calculations that correctly escalate triage levels (e.g., vital sign modifiers for ESI Level 3 patients).
• Formatting complex medical notes into standardized hospital SBAR and SOAP conventions ready for EHR copy-pasting.

🏆 Accomplishments That We're Proud Of

• Created an end-to-end working system that bridges patient intake, clinical evaluation, and provider queue management in one unified interface.
• Zero-configuration functionality: anyone can test the system immediately with built-in clinical benchmark cases without needing an external API key or setup.
• Clean, accessible, and responsive UI suitable for hospital tablets and desktop workstations.

📚 What We Learned

• Deepened our understanding of clinical triage frameworks (ESI v4, AVPU sensorium scale, SIRS sepsis criteria).
• Learned how to structure and engineer prompts and deterministic logic for high-stakes healthcare workflows.

🔮 What's Next for MedAssist Triage

• Direct EHR / FHIR API Integration: Seamless sync with hospital electronic health record systems (Epic, Cerner).
• Multilingual Voice Intake: Expanding speech-to-text intake in 50+ local languages for global emergency centers.
• Wearable & Bedside IoT Telemetry Stream: Real-time Bluetooth vital sign streaming from connected pulse oximeters and monitors.

Built With

• artifical-intelligent
• healthcare
• react
• tailwindcss
• typescript
• vite