AI Services

Blood Connect runs AI workloads as Python FastAPI services under infrastructure/ai-services/, wired into the Next.js app via NEXT_PUBLIC_AI_SERVICES_URL and tRPC routers such as matching.

Available

Donor–Recipient Matching

Ranks eligible donors for a blood request using blood type compatibility, proximity, availability, and urgency.

•Blood type compatibility
•Geographic proximity
•Availability and fatigue signals
•Urgency prioritization
•Exposed via tRPC matching router and FastAPI service

Available

Demand Forecasting

Predicts blood demand patterns to support inventory planning and donor recruitment.

•Demand prediction models
•Inventory planning support
•Hospital / blood-bank analytics inputs
•Stored as DemandPrediction records

Available

Risk Assessment

Supports donor eligibility and donation-safety evaluation using structured health and history signals.

•Eligibility-oriented risk scoring
•WHO-aligned screening context
•FastAPI risk assessment API
•Used alongside digital health screening flows

How services are deployed

The unified FastAPI entrypoint is infrastructure/ai-services/main.py. Matching has a dedicated module at infrastructure/ai-services/matching/; forecasting and risk assessment live alongside shared middleware, caching, and monitoring.

The web app does not ship a public npm/Python SDK. Integrate through the Blood Connect backend (tRPC / internal AI client), not a direct third-party SDK package.

Application integration

Typical product paths:

Hospital / emergency blood requests → matching via bloodRequests + matching routers
Inventory planning → forecasting outputs / analytics
Donor screening → risk assessment + WHO screening flows

See Developer API

Note: There is no production AI chatbot, public AI playground, or published @bloodlink/ai-sdk package in this repository. Accuracy percentages shown in older marketing copy are not product SLAs.