-
-
CardioTwin AI v2: Live Clinical Dashboard & Predictive AI Warning Interface.
Inspiration
Cardiovascular diseases remain the leading cause of death globally. Most medical responses occur after a crisis has already happened, when time is severely limited. We were inspired to build CardioTwin AI v2 to shift this paradigm from reactive management to proactive prevention, utilizing artificial intelligence to detect dangerous trends before physical symptoms manifest.
What it does
CardioTwin AI v2 acts as a continuous digital twin for high-risk patients. By ingestion and analysis of real-time sequential time-series biometric streams (Heart Rate, HRV, and Blood Oxygenation) from simulated consumer wearables, the system predicts sudden cardiac arrests and vascular anomalies up to 45–60 minutes in advance. If an upcoming crisis profile is calculated with >91% accuracy, the platform automatically triggers emergency protocols, broadcasting live patient coordinates to medical triage facilities.
How we built it
The application core engine is entirely built using Python to implement the predictive logic and handle data processing. The user interface was developed using Streamlit, establishing a responsive, dynamic clinical dashboard that simulates real-time data stream tracking and provides distinct, automated red-alert visual states when a medical anomaly is actively identified.
Challenges we ran into
A major technical challenge was simulating sequential time-series biometric data that accurately mimics complex clinical telemetry—like sudden heart rate spikes paired with oxygen saturation drops—without active medical sensors attached during the build. We solved this by programming a dynamic data generator that perfectly matches genuine clinical desaturation profiles.
Accomplishments that we're proud of
We successfully designed and ran a fully operational working prototype that addresses a critical global health problem within an optimized software architecture. We are incredibly proud to present a scalable solution aimed at directly saving human lives.
What we learned
We gained deep technical insights into time-series predictive modeling and UI state management using Python. We also learned how critical immediate, automated, and encrypted data routing is when handling real-time healthcare systems where seconds matter.
What's next for CardioTwin AI v2
We intend to integrate real Bluetooth telemetry inputs from physical smartwatches and validate our predictive core model against open-source clinical heart disease datasets to further enhance prediction accuracy and scalability.
Built With
- predictive
- python
- streamlit
