🌤️ Building an Explainable Weather–Health Risk Intelligence System Using OpenWeather, FastAPI, and Local LLMs

#climatechange #healthtech #ai #climatetech

🚀 I’m excited to share a project I’ve been developing at the intersection of climate resilience, public health, and explainable AI.

I recently built and released an Explainable Weather–Health Risk Intelligence System that uses OpenWeather, FastAPI, and a local LLM (Ollama) to translate real-time weather and air-quality data into clear, personalized health-risk insights.

As climate-driven heat waves, pollution spikes, and humidity extremes increase across the U.S., many communities—especially vulnerable groups—lack accessible tools to interpret how daily environmental conditions affect their health. Traditional weather apps rarely communicate risk in a medically relevant or explainable way.

This project explores how we can bridge that gap using lightweight, local-first AI and transparent risk-scoring models.

🌍 What the System Contributes
Using real-time data from OpenWeather, the system converts:

Temperature
Humidity
AQI
PM2.5
PM10
O₃

Into four core risk metrics:

Asthma / respiratory risk
Heat-stress risk
Dehydration risk
Overall daily health-risk index

It also generates a 24-hour forecast-based risk outlook, helping users understand how risk changes throughout the day.

🧠 Local LLM Explainability
Using a local LLM running through Ollama, the system produces:

Clear explanations
Personalized safety recommendations
Profile-specific notes (age + underlying conditions)

This keeps data private and enables fully offline deployments — ideal for public-health environments or low-resource settings.

🧩 Technical Foundations
FastAPI (Python) backend:

React + TypeScript + Tailwind frontend
Ollama for local LLM inference (no cloud model required)
OpenWeather APIs for weather + air-quality retrieval

The architecture is designed for scalability, local privacy, and smooth integration into health-analytics workflows, climate-health dashboards, or community-resilience tools.

🧪 Code Repositories

Backend (FastAPI):
🔗 https://github.com/ddake25/openweatherchallengeBE
Frontend (React)
🔗 https://github.com/ddake25/openweatherchallengeFE

Both are fully open-source and available for developers, researchers, and public-health practitioners who want to build on top of the framework.

🎯 Why This Matters
This project strengthens my ongoing work in combining:

AI,
environmental intelligence, and
health literacy, — an area increasingly recognized as critical to U.S. public health, emergency preparedness, and climate-adaptation planning.

Improving how everyday people interpret environmental risk—especially those with asthma, heat sensitivity, or dehydration risk—supports broader national goals around: