SUS-Twin
SUS-Twin uses IoT and AI to eliminate medicine shortages across Brazilian cities through predictive logistics and collaborative stock sharing.
Project Description
- Proposal Details & Overview
SUS-Twin is a Digital Twin platform designed to solve Brazil’s chronic medicine shortages, which affect 80% of cities and so far result in R$2.3B waste. Focusing on Health and Sustainability, our solution transforms reactive management into a predictive, circular ecosystem.
Creativity & Innovation:
Unlike basic inventory systems, SUS-Twin uses AI to forecast demand based on seasonal trends and IoT for real-time thermal monitoring of critical assets. We introduced “Solidarity Logistics,” an automated system that redistributes surplus stock between municipalities before expiration.
Social Impact:
We ensure that life-saving treatments reach the families who need them most. By reducing emergency purchases (50% more expensive) and preventing vaccine loss due to cooling failures, we protect public funds and lives, directly impacting SDGs 3, 9, 12, and 17.
Technical Feasibility & Collaboration:
The AI4Humanity team integrated Product and Engineering roles to build a functional prototype. Our AI-accelerated workflow allowed us to transform complex logistics data into an intuitive dashboard and scalable infrastructure in record time.
- Technologies, Frameworks, and AI Tools
Our technical execution leveraged a high-performance AI ecosystem to ensure precision and speed:
AI & Research: Google Gemini 2.5 Flash served as our core engine for strategic research, demand forecasting, and OCR logic for expiration date scanning. We also integrated the Twin_one ML service, a dedicated microservice for advanced seasonality analysis.
Development (Back-end): Coding was streamlined using Node.js and Python (FastAPI), accelerated by Cursor, Claude, and ChatGPT to ensure rapid deployment and clean architecture.
Security & Authentication: Implemented JWT (JSON Web Tokens) for robust access control, ensuring secure data exchange between health units and the centralized management portal.
Execution & Testing: Google Colab was utilized for developing and validating AI algorithms, providing a collaborative environment for data integration.
Technology Stack:
Framework: FastAPI and Node.js for high-speed, low-latency communication with IoT sensors and external APIs.
Database: MongoDB Atlas, chosen for its high availability and flexibility in handling unstructured inventory and telemetry data.
Integration: Designed for seamless connection with the National Health Data Network (RNDS).
Responsible Application:
These tools were applied to maintain public data sovereignty and provide evidence-based insights, ensuring technology empowers healthcare managers to make life-saving decisions.