Spag | Devpost

Inspiration

We were inspired by the "Invisible Crisis" of urban health in Lagos. While our devices can map the entire world, they remain silent as we face 38°C heat and five-hour traffic jams on the 3rd Mainland Bridge. We realized that while we over-collect health data, we fail to act before a biological "crash" occurs. S.P.A.G. was born to bridge this gap—moving healthcare from reactive observation to proactive, moment-by-moment prevention.

What it does

S.P.A.G. is a Preventive Health Operating System that transforms passive signals into predictive interventions. It calculates a dynamic Health State Vector to forecast needs:

Hunger Timer: Predicts the next hunger window based on meal size, activity, and historical rhythms.
Hydration Score: A 0–100 score that factors in local weather, activity levels, and water intake.
Multilingual Nudges: Uses AI to send personalized alerts in Pidgin, Yoruba, Hausa, or Igbo.
Universal Design: A voice-first interface that ensures health tracking is inclusive for those with visual or physical accessibility needs.

How we built it

We architected a Tri-Pillar Grid Model for low-latency resilience:

Frontend: Built with React Native and Expo for real-time sensor sync and lightweight logging.
Backend: A Python + FastAPI engine managing time-series regression and LLM orchestration.
Database: PostgreSQL acting as the long-term memory of the system.
Visualization: A Next.js 16 dashboard using Three.js to render a 3D Composite Twin of the user’s physiological state.

Challenges we ran into

Cultural Context: Training models to recognize metabolic differences between Nigerian staples like Pounded Yam and Tuwo Shinkafa.
Edge Inference: Optimizing AI reasoning to function in low-bandwidth environments with partial offline capability.
Sovereign Security: Implementing local encryption and user-owned keys so health data remains a private “map of the soul.”

Accomplishments that we're proud of

Inclusivity: Integrating 250+ ethnic contexts and a voice-first interface that empowers users regardless of physical ability.
The Feedback Loop: Developing a self-correcting model that learns from every log and continuously improves prediction accuracy.
Infrastructure Thinking: Moving beyond a simple mobile app into a scalable, clinical-grade preventive health infrastructure.

What we learned

The most critical component of an AI system is the human feedback loop. Building for the extreme conditions of Lagos—heat, traffic, and noise—revealed that health tracking must be frictionless. Every extra second of effort can reduce user compliance by ~20%.

We also learned that data sovereignty is foundational to trust. Users must control their health data for digital health systems to gain real adoption.

What's next for Spag

Our roadmap focuses on scaling human resilience while maintaining low latency and strong privacy guarantees:

Edge Model Deployment: Delivering efficient localized models capable of offline reasoning.
Wearable Expansion: Integrating standardized SDKs for ECG sensors and continuous hydration tracking devices.
Clinical Integration: Partnering with preventive care clinics for validation studies and Electronic Health Record (EHR) interoperability.