Inspiration
Inspired by the shift in healthcare from reactive treatment to proactive prevention, our goal was to bridge the gap between mental and physical wellness by creating an integrated "digital twin" that makes personalized coaching accessible to everyone.
What it does
Cadence AI is an intelligent health assistant that uses conversational onboarding to generate personalized workout schedules and track mental recovery through journal sentiment analysis. It calculates a holistic fitness score by combining cardiovascular biomarkers with machine learning predictions trained on thousands of fitness-related data points.
How we built it
The platform features a React (Vite) frontend with a glassmorphism design and a FastAPI backend for high-performance data processing. We utilized Snowflake Cortex AI and Gemini for enterprise-grade LLM tasks and Scikit-learn to build a hybrid ML model for fitness scoring.
Challenges we ran into
One of our biggest hurdles was converting unstructured conversational data into structured JSON for our data warehouse without losing context.
Accomplishments that we're proud of
We successfully replaced tedious registration forms with a seamless, three-question AI-led conversation. We are also proud of achieving minimal latency for sentiment analysis by leveraging SQL-based AI functions directly within the Snowflake environment.
What we learned
Building this platform taught us that mental sentiment is a powerful leading indicator for physical performance and recovery. We also gained deep insights into "Healthcare Reengineering," specifically how AI can democratize access to expertise that was previously gated by high costs.
What's next for Cadence AI
Our next steps could be setting up SMS notifications, where a user receives reminders to work out, fill out, complete their journal entry, etc. Another area to consider is integrating wearable data from Apple Watch and Fitbit to automate the tracking of heart rate variability and sleep stages, or develop "Telehealth Triggers" that can flag high-distress language for early intervention by licensed professionals. The possibilities are limitless.
Log in or sign up for Devpost to join the conversation.