Kids Outbreak Radar

Nova agents turn complex outbreak data into one simple answer for parents: is it safe to send my child to school today?

Comment

Using Amazon Nova

Outbreak Radar for Kids is a mobile health application that helps parents protect their children from disease outbreaks through AI-powered, personalized recommendations. The app leverages Amazon Bedrock Nova foundation models as its core intelligence engine to analyze CDC outbreak data and generate age-appropriate health guidance.

The application uses a sophisticated dual-model strategy with Amazon Nova Lite and Nova Micro. Nova Lite handles complex outbreak scenarios involving multiple high-severity diseases or elevated risk levels, providing detailed analysis and comprehensive recommendations. Nova Micro processes simpler scenarios with low risk or single-disease outbreaks, optimizing for cost efficiency. This intelligent model selection is implemented through a custom ModelSelector component that evaluates outbreak complexity based on disease count and severity variance.

Nova models receive structured prompts containing the child's age range (infant, toddler, preschool, or school-age), geographic location, current outbreak data with severity scores, and language preference (English or Japanese). The models generate 3-5 actionable recommendations such as "monitor temperature daily," "practice thorough handwashing," or "consult healthcare provider if symptoms appear." Each recommendation is validated for safety, relevance, and age-appropriateness before being presented to parents.

The system achieves remarkable cost efficiency at under $0.0001 per request through 24-hour recommendation caching, background pre-generation for common profiles, and the dual-model strategy. Response times remain under one second, ensuring a smooth user experience. Built on AWS serverless architecture with Lambda, DynamoDB, and EventBridge, the app automatically updates CDC data weekly and provides an interactive map interface for visualizing outbreak patterns across regions.

Inspiration

Parents struggle to track multiple disease outbreaks affecting their children. I built an AI-powered early warning system that provides personalized, actionable health guidance based on real-time outbreak data and child age groups.

What it does

Kids Outbreak Radar monitors infectious disease outbreaks (flu, RSV, COVID-19, etc.) across regions and generates age-appropriate prevention recommendations using Amazon Bedrock's Nova AI. Parents receive real-time alerts and personalized guidance tailored to their children's ages and locations.

How we built it

We built a React Native mobile app with AWS serverless backend (Lambda, DynamoDB, API Gateway). The outbreak-fetcher Lambda aggregates data from CDC and NIID APIs, while Nova AI generates contextual recommendations with Bedrock Guardrails ensuring child-safe content.

Challenges we ran into

Integrating multiple public health APIs with inconsistent data formats was challenging. We also had to optimize Nova AI prompts to generate age-appropriate, actionable advice while maintaining medical accuracy and avoiding alarming language for parents.

Accomplishments that we're proud of

We successfully created a production-ready system that processes real outbreak data and generates personalized recommendations in under 2 seconds. Our 24-hour caching strategy and background pre-generation ensure instant app launches while minimizing AI costs.

What we learned

We learned how to effectively combine multiple AWS services (Bedrock, Lambda, DynamoDB) to create a responsive AI application. We also gained insights into designing child-focused UX and balancing information delivery without causing parental anxiety.

What's next for Kids Outbreak Radar

We plan to add push notifications for high-risk outbreaks, expand to more countries, integrate school closure data, and implement community-driven outbreak reporting. We'll also add multilingual support and accessibility features for diverse families.

Share this project:

Updates