LifeLine

Inspiration

Mediterranean crises often fail not because of lack of doctors, but because hospitals get overloaded before coordination catches up. We wanted to build a system-level early warning tool that helps responders act 24-48 hours earlier, not a generic chatbot.

What it does

CrisisLoad AI predicts hospital overload risk using operational signals (bed/ICU occupancy, ER inflow, ambulance arrivals, supply burn, infrastructure risk, staffing).
It gives:

Overload probability
Green/Amber/Red risk status
Estimated time-to-overload
Triage scoring for incoming patients
Resource/referral guidance
AI-generated action plans for responders

How we built it

We built a full local deployable web system:

Backend: Node.js + Express + TypeScript, secure auth (JWT + hashed passwords), country-aware access control
Intelligence layer: weighted risk model + triage engine + referral logic
Frontend: React + TypeScript responsive dashboard (mobile + laptop)
AI layer: Gemini-assisted operational recommendations
Data: realistic synthetic crisis/hospital simulation

Risk core uses a logistic-style score: [ P(\text{overload})=\sigma\left(w_1x_1+w_2x_2+\cdots+w_nx_n\right)\times 100 ]

Challenges we ran into

No real hospital data, so we had to design believable synthetic operational data
Balancing speed vs. production structure under tight time
Handling local deployment constraints and CORS/auth issues across dynamic localhost ports
Keeping outputs operationally useful, not just “AI-sounding”

Accomplishments that we're proud of

Delivered a working end-to-end product, not just a mockup
Built country-scoped responder workflows (Lebanon, Palestine, Turkiye, Italy, Spain)
Made the platform both decision-support and demo-ready
Kept the model explainable and practical for emergency teams

What we learned

In emergencies, clarity beats complexity
Explainable models are easier to trust for frontline decisions
Strong UX and operational framing matter as much as model accuracy
Production thinking (auth, roles, deployment, reliability) is a major differentiator